Resources

Enigma germline classification

How Enigma Clinical classifies germline small variants automatically

28
ACMG/AMP criteria: 8
computed automatically: 2
raised for curator review: 28of 123 released
ClinGen gene specifications applied

Overview

Every single-nucleotide variant and small indel page in Enigma Clinical (GRCh38, alleles up to 50 bases, nuclear genome) carries an automated germline classification. It follows the ACMG/AMP 2015 standard (Richards et al. 2015(opens in a new tab)) as refined by the ClinGen Sequence Variant Interpretation working group, and adds up the evidence with ClinGen’s points system (Tavtigian et al. 2020(opens in a new tab)). The scoring engine is ENIGMA Variant Classification v1.0, the points engine of Enigma Genomics’s variant reporting system.

The result is for research use. It is built only from public data: it never sees the patient, the family, the phenotype or laboratory studies, which carry much of the weight in a clinical classification. Criteria that need such evidence are listed on every variant page as not evaluated, and the criteria the data only point to are raised for a curator instead of being scored.

What happens for each variant

Evidence is collected from public sources in parallel: frequencies and coverage from gnomAD, the consequence on the MANE Select transcript with LOFTEE from Ensembl VEP, a calibrated missense score from dbNSFP, classified variants at the same residue from ClinVar, and gene-level dosage and validity from ClinGen and GenCC. A source that does not answer marks its criteria as not evaluated instead of failing the whole result.
ClinGen gene-specific rules are looked up. When a ClinGen Variant Curation Expert Panel has published a specification for the gene and the engine implements it, its thresholds and domain definitions replace the generic ones and the codes it does not use are marked not applicable. When a specification exists but is not implemented yet, the page says so.
Each of the 28 criteria gets a status (met, for review, not met, not evaluated or not applicable), a strength and the evidence behind it, grouped by evidence category.
Met criteria are scored on the points scale, exclusions and caps are applied, and the total falls into one of five classes.
A ClinGen expert-panel classification of the same variant, when one exists in the ClinGen Evidence Repository, is shown above the automated result and takes precedence over it.

Points and classes

Pathogenic criteria add points and benign criteria subtract them; the size depends on the strength. BA1 is stand-alone: when it is met the variant is Benign whatever else applies.

Points by strength

Supporting: ±1
Moderate: ±2
Strong: ±4
Very Strong: ±8

Classes by total

Pathogenic: ≥ 10
Likely Pathogenic: 6 to 9
Uncertain Significance: 0 to 5
Likely Benign: −1 to −6
Benign: ≤ −7

Rules applied while scoring

When PVS1 is met, PP3, PM4 and BP4 are removed (double count: PVS1 already covers this loss-of-function evidence).
When PM1 is met, PP2 and BP1 are removed (excluded when PM1 applies).
When PP2 is met, BP1 is removed (excluded when PP2 applies).
When BS2 is met, PM2 is removed (BS2 and PM2 are mutually exclusive).
When PS2 is met, PM6 is removed (de novo cases combine on one PS2 points scale).
BA1 removes BS1.
PM1 counts at most Moderate, and PM1 with PP3 at most +4 together (Pejaver et al. 2022(opens in a new tab)).
PP5 and BP6 are never scored: retired by ClinGen SVI 2018: another lab's assertion is not evidence (Biesecker and Harrison 2018(opens in a new tab)).
PS3 and BS3 are curator-only: functional evidence needs a curator to judge the assay (Brnich et al. 2020(opens in a new tab)).
A Pathogenic or Likely Pathogenic total is capped at Uncertain Significance when the gene’s best gene–disease validity is below Moderate (Definitive, Strong, Moderate allow it; ClinGen where curated, otherwise GenCC).
Pathogenic evidence of at least +4 against benign evidence of at least 4 points is flagged as conflicting for curator review.
A ClinGen gene specification may use two codes together that the generic exclusions separate (for example PM1 with PP2).

Criteria and automated rules

The generic rule for every criterion, with its data source. Gene-specific rules are in the next sections.

Criteria are grouped as in the ACMG/AMP evidence framework: population data, effect on protein and computational evidence, functional data, segregation, de novo, allelic data, phenotype, other. “Raised for review” criteria are shown on the variant page when the data point to them but never add points; “curator only” criteria need case, family, phenotype or laboratory evidence that public data cannot supply.

28 rows

The 28 ACMG/AMP criteria and how the engine evaluates each one
Code	Evidence category	Criterion	Automation	Automated rule	Data source
`PVS1`	Effect on protein and computational evidence	Null variant in a gene where loss of function causes the disease.	Automated	Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.	Ensembl VEP (LOFTEE); ClinGen Dosage Sensitivity
`PS1`	Effect on protein and computational evidence	Same amino-acid change as an established pathogenic variant.	Automated	Same amino-acid change from a different nucleotide change, classified in ClinVar with at least 2 stars: Pathogenic gives Strong, Likely pathogenic gives Moderate. Fewer stars: for review.	ClinVar (NCBI E-utilities)
`PS2`	De novo	De novo, with maternity and paternity confirmed.	Curator only	De novo status needs parental testing.	Curator
`PS3`	Functional data	Well-established functional studies show a damaging effect.	Curator only	Functional studies are curator-only (Brnich et al. 2020).	Curator
`PS4`	Population data	Significantly more frequent in affected individuals than in controls.	Curator only	Needs case–control data or counts of affected probands.	Curator

Thresholds and their sources

Frequencies use the gnomAD v4.1 joint exome and genome data: the filtering allele frequency (FAF95) of the genetic-ancestry group with the highest value, counted only in groups with at least 2,000 alleles for BA1 (Ghosh et al. 2018(opens in a new tab)). Absence from gnomAD supports PM2 only where the mean depth is at least 20×. PP3 and BP4 use one tool fixed in advance, never several: AlphaMissense, or BayesDel without allele frequency when AlphaMissense has no score, at the ClinGen-calibrated thresholds (Bergquist et al. 2025(opens in a new tab); Pejaver et al. 2022(opens in a new tab)). PS1 and PM5 compare against ClinVar variants with at least 2 review stars. PVS1 follows a simplified ClinGen decision tree on LOFTEE calls (Abou Tayoun et al. 2018(opens in a new tab)).

17 rows

Generic thresholds and their sources
Criterion	Parameter	Value	Source
`BA1`	grpmax FAF95 above	`5%`	Ghosh et al. 2018(opens in a new tab)
`BA1`	Minimum alleles in the group	`2,000`	Ghosh et al. 2018(opens in a new tab)
`BS1`	Review when grpmax FAF95 above	`1%`	Enigma policy: BS1 is disease-specific
`PM2`	Allele frequency below (no homozygotes)	`0.001%`	ClinGen SVI (PM2 at Supporting)(opens in a new tab)
`PM2`	Mean depth for absence to count	`20×`	Enigma policy

Gene-specific rules (ClinGen CSpec)

ClinGen Variant Curation Expert Panels (VCEPs) adapt the ACMG/AMP criteria to single genes and diseases and publish the result in the ClinGen Criteria Specification Registry (CSpec). The registry lists 211 specifications, of which 123 are released, covering 134 genes (registry read 2026-10-25). A copy of the registry index and of each implemented specification is kept in Enigma Clinical’s knowledge base, so a result can always be traced to the exact version used.

Implemented specification: its thresholds, domains and strength limits replace the generic rules for that gene, the codes it does not use are shown as not applicable, and the variant page says “ClinGen gene-specific guidelines apply” with the panel and version.
Released but not implemented: the generic rules apply and the variant page names the specification, because its thresholds can differ and it applies only to the disease it was written for (a specification for beta-thalassaemia, for example, says nothing about sickle cell disease).
Tools the panel names that are not licence-free (REVEL, CADD, SpliceAI) are never used. Where a panel specifies one, the engine uses its own calibrated predictor and keeps the panel’s strength limit; the variant page says so in the rule.
Case-level criteria stay with curators even when a panel defines them precisely (de novo points, segregation meioses, case–control points, functional assays).

MYH7: Cardiomyopathy VCEP, version 2.0 (GN002)

ClinGen Cardiomyopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MYH7 Version 2.0. Released, last updated 2024-04-22; disease dilated cardiomyopathy (MONDO:0005021, autosomal dominant inheritance); hypertrophic cardiomyopathy (MONDO:0005045, autosomal dominant inheritance). Residues and exons are read on NM_000257.4 (ENST00000355349.4). Specification GN002(opens in a new tab)

28 rows

ClinGen Cardiomyopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MYH7 Version 2.0: criteria and how the engine applies them
Code	In the engine	Cardiomyopathy VCEP rule (v2.0)	How the engine applies it
`PVS1`	Not applicable	Not applicable for MYH7 (Cardiomyopathy VCEP v2.0).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: No cardiomyopathy specifications. Apply as outlined by Richards et al. 2015[<sup>1</sup>](#pmid_25741868). Example of when rule should NOT be applied. NM_000256.3(MYBPC3): c.2308G>A (p.Asp770Asn) has an established impact on splicing leading to nonsense mediated decay (NMD) and should not be used to provide evidence for other variants observed to result in the same amino acid change.	PS1 as in Richards et al. 2015 (no cardiomyopathy change).
`PS2`	Curator only	Strong: Refer to SVI guidance on number/combination of cases required based on phenotype specificity[<sup>2</sup>](#url_c73e109e-b916-5a72-b7b1-1762446f3c11)). For most cardiomyopathies, it is recommended to default to Phenotype consistency: “Phenotype consistent with gene but not highly specific”. Clinical judgment is required for shifting to a higher or lower category. For use as a STRONG or VERY STRONG criterion, ideally parents have been thoroughly clinically evaluated without evidence of cardiomyopathy (ideally using a combination of ECG and echocardiogram or cardiac MRI for maximum sensitivity). A family history consistent with de novo inheritance should not have any clinical signs or symptoms suggestive of cardiomyopathy in a 1<sup>st</sup> or 2<sup>nd</sup> degree relative, for example: 1. Sudden death under 60 years of age 2. Heart transplant 3. Implantable cardiac defibrillator (ICD) under 60 years of age 4. Features of cardiomyopathy (e.g., systolic dysfunction, hypertrophy, left ventricular enlargement in an individual without risk factors). 5. Other related/overlapping cardiomyopathies Examples of non-suspicious family history may include non-specific clinical features (e.g., palpitations, syncope, borderline/inconclusive echocardiogram findings, heart attack if age appropriate and suspected to result from coronary artery disease), but every attempt should be made to clarify features. Generally, this criterion is only applicable in the ABSENCE of any other possible disease-causing variants. If other pathogenic or likely pathogenic variants are present, consider decreasing points assigned or overall weight.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: In vitro splicing assays (e.g., RNA studies) In vitro splicing assays may be considered as STRONG evidence, providing the following criteria are met. * Prior knowledge of predominant transcripts in cardiac tissue Analysis undertaken using RNA extracted from cardiac tissue from the individual with the variant Analysis undertaken using RNA extracted from whole blood providing the relevant transcripts (isoforms) are expressed in blood and are at sufficient levels to assess splice disruption. Assay shows a clear, reproducible and convincing effect on splicing (i.e. a distinct splice product, present at a level comparable to the splice product from the wild-type allele), which is not observed in controls * Confirmation of abnormal splice product by Sanger sequencing NOTE: Mini-gene assay in non-patient derived cell lines are NOT considered to provide STRONG evidence. NOTE: Whether to activate this rule needs to be reconciled with the variant spectrum and disease mechanism for the gene at hand (i.e., consider whether the effect is likely to lead to LOF or an in-frame alteration and whether this type of effect is expected to be disease causing) (Abou Tayoun et al. 2018[<sup>3</sup>](#pmid_30192042)). · Moderate: In vivo models (e.g., variant knock-in animal models) Mammalian variant-specific knock-in animal models that produce a phenotype consistent with the clinical phenotype in humans (e.g., structural and/or functional cardiac abnormalities, premature death, arrhythmia) may be considered as MODERATE evidence NOTE: The following assays/models do NOT meet criteria 1. Assays that are known to be associated with non-specific cardiac phenotypes (e.g., morpholino-induced pericardial edema in zebrafish) 2. In vivo evidence that is not variant specific, such as whole gene alterations (i.e., cDNA or whole gene transgenic mice and whole or partial gene knock-out mice) · Supporting: In vitro assays (e.g., biochemical assays of myofilament function, motility assays, human iPSC-CM) While some in vitro assays may provide evidence that a variant in a cardiomyopathy gene has an effect on protein and/or myofilament function, at present, there are no validated “gold-standard” assays that are considered to reliably predict the clinical phenotype. As such, in the cardiomyopathy genes listed in these guidelines, data from individual in vitro studies are unlikely to meet the criteria required to assign this rule at more than SUPPORTING level.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Currently few well-designed case-control studies have been performed for inherited cardiomyopathies. Until such studies become available, comparative analyses can be undertaken using case data (e.g., internal and/or published cohorts) and control data from population-level cohorts (e.g., gnomAD). Cohorts used in these analyses should meet the following criteria: 1. The cases have a clinical diagnosis of the specified disorder or related phenotype (e.g., all cases have HCM or another relevant phenotype). When assessing cases, it's important to consider how likely another potential cause of the phenotype has been excluded. This includes considering the presence of other variants in relevant genes (particularly those likely to be contributing to phenotype) and the extent of testing performed (i.e., single gene sequencing, panel testing, whole exome/genome sequencing). 2. The controls should not be derived from study populations that might be enriched for the specified disorder. 3. The denominator of the cohorts must be available (e.g., variant detected in 5 out of 3,500 cases and 1 out of 60,000 controls). 4. The cohorts do not include closely related individuals (i.e., family members are not included in the case counts). 5. The cohorts do not overlap with other cohorts being used in the analysis (i.e., cases are not being counted more than once). 6. The population diversity of the case and control cohorts are broadly similar. 7. Consider the size of the case cohort — larger cohorts are likely to provide more accurate estimates of variant frequency; therefore, it may be preferable to use data from the largest available case series for case-control analyses (e.g., Walsh et al. 2017[<sup>8</sup>](#pmid_27532257), [DECIPHER](https://www.deciphergenomics.org/gene/MYH7/patient-overlap/snvs)). To account for limitations that arise when performing unmatched case-control analyses, the following stringent OR threshold is recommended: * STRONG evidence requires the lower bound of the 95% confidence interval (CI) around the odds ratio (OR) estimate to be ≥20 A PS4 calculator is available at [www.cardiodb.org](https://www.cardiodb.org/ps4_calculator/ps4_calculator.html). If multiple cohorts are available, the final ORs and associated CIs need to be harmonized across all cohorts to determine the final level (e.g., if 2 large cohorts have an OR of ~6 and a third small cohort has an OR of 11, application at a SUPPORTING level should be considered). *RELEVANT PHENOTYPES: 1. Cases of HCM and RCM may be combined as they are considered part of the same disease spectrum. 2. For the eight genes covered by these guidelines, the combination of probands with other phenotypes should be reviewed by a clinical expert to determine if grouping is appropriate. 3. Additional considerations for LVNC and end-stage HCM: * Due to the current debate about whether isolated LVNC represents a true disease entity or variation of typical cardiac morphology (Anderson et al. 2017[<sup>9</sup>](#pmid_28395867); Oechslin et al. 2017[<sup>6</sup>](#pmid_28545618); Hershberger et al. 2017[<sup>5</sup>](#pmid_29212902); Ross et al. 2020[<sup>7</sup>](#pmid_31143950)), individuals with isolated LVNC should NOT be added to proband or segregation counts (including individuals with isolated LVNC in a family with other cardiomyopathies). HCM and DCM have distinct mechanisms of disease and therefore pathogenetic variants are not anticipated to cause both primary phenotypes. While occurrence in both phenotypes may initially be considered as evidence against pathogenicity, end-stage HCM can present similarly to DCM. Careful consideration is needed before including DCM or related phenotypes in case or segregation data for primarily HCM variants. · Moderate: Currently few well-designed case-control studies have been performed for inherited cardiomyopathies. Until such studies become available, comparative analyses can be undertaken using case data (e.g., internal and/or published cohorts) and control data from population-level cohorts (e.g., gnomAD). Cohorts used in these analyses should meet the following criteria: 1. The cases have a clinical diagnosis of the specified disorder or related phenotype (e.g., all cases have HCM or another relevant phenotype). When assessing cases, it's important to consider how likely another potential cause of the phenotype has been excluded. This includes considering the presence of other variants in relevant genes (particularly those likely to be contributing to phenotype) and the extent of testing performed (i.e., single gene sequencing, panel testing, whole exome/genome sequencing). 2. The controls should not be derived from study populations that might be enriched for the specified disorder. 3. The denominator of the cohorts must be available (e.g., variant detected in 5 out of 3,500 cases and 1 out of 60,000 controls). 4. The cohorts do not include closely related individuals (i.e., family members are not included in the case counts). 5. The cohorts do not overlap with other cohorts being used in the analysis (i.e., cases are not being counted more than once). 6. The population diversity of the case and control cohorts are broadly similar. 7. Consider the size of the case cohort — larger cohorts are likely to provide more accurate estimates of variant frequency; therefore, it may be preferable to use data from the largest available case series for case-control analyses (e.g., Walsh et al. 2017[<sup>8</sup>](#pmid_27532257), [DECIPHER](https://www.deciphergenomics.org/gene/MYH7/patient-overlap/snvs)). To account for limitations that arise when performing unmatched case-control analyses, the following stringent OR threshold is recommended: * MODERATE evidence requires the lower bound of the 95% CI around the OR to be ≥10 A PS4 calculator is available at [www.cardiodb.org](https://www.cardiodb.org/ps4_calculator/ps4_calculator.html). If multiple cohorts are available, the final ORs and associated CIs need to be harmonized across all cohorts to determine the final level (e.g., if 2 large cohorts have an OR of ~6 and a third small cohort has an OR of 11, application at a SUPPORTING level should be considered). *RELEVANT PHENOTYPES: 1. Cases of HCM and RCM may be combined as they are considered part of the same disease spectrum. 2. For the eight genes covered by these guidelines, the combination of probands with other phenotypes should be reviewed by a clinical expert to determine if grouping is appropriate. 3. Additional considerations for LVNC and end-stage HCM: * Due to the current debate about whether isolated LVNC represents a true disease entity or variation of typical cardiac morphology (Anderson et al. 2017[<sup>9</sup>](#pmid_28395867); Oechslin et al. 2017[<sup>6</sup>](#pmid_28545618); Hershberger et al. 2017[<sup>5</sup>](#pmid_29212902); Ross et al. 2020[<sup>7</sup>](#pmid_31143950)), individuals with isolated LVNC should NOT be added to proband or segregation counts (including individuals with isolated LVNC in a family with other cardiomyopathies). HCM and DCM have distinct mechanisms of disease and therefore pathogenetic variants are not anticipated to cause both primary phenotypes. While occurrence in both phenotypes may initially be considered as evidence against pathogenicity, end-stage HCM can present similarly to DCM. Careful consideration is needed before including DCM or related phenotypes in case or segregation data for primarily HCM variants. · Supporting: Currently few well-designed case-control studies have been performed for inherited cardiomyopathies. Until such studies become available, comparative analyses can be undertaken using case data (e.g., internal and/or published cohorts) and control data from population-level cohorts (e.g., gnomAD). Cohorts used in these analyses should meet the following criteria: 1. The cases have a clinical diagnosis of the specified disorder or related phenotype (e.g., all cases have HCM or another relevant phenotype). When assessing cases, it's important to consider how likely another potential cause of the phenotype has been excluded. This includes considering the presence of other variants in relevant genes (particularly those likely to be contributing to phenotype) and the extent of testing performed (i.e., single gene sequencing, panel testing, whole exome/genome sequencing). 2. The controls should not be derived from study populations that might be enriched for the specified disorder. 3. The denominator of the cohorts must be available (e.g., variant detected in 5 out of 3,500 cases and 1 out of 60,000 controls). 4. The cohorts do not include closely related individuals (i.e., family members are not included in the case counts). 5. The cohorts do not overlap with other cohorts being used in the analysis (i.e., cases are not being counted more than once). 6. The population diversity of the case and control cohorts are broadly similar. 7. Consider the size of the case cohort — larger cohorts are likely to provide more accurate estimates of variant frequency; therefore, it may be preferable to use data from the largest available case series for case-control analyses (e.g., Walsh et al. 2017[<sup>8</sup>](#pmid_27532257), [DECIPHER](https://www.deciphergenomics.org/gene/MYH7/patient-overlap/snvs)). To account for limitations that arise when performing unmatched case-control analyses, the following stringent OR threshold is recommended: * SUPPORTING evidence requires the lower bound of the 95% CI around the OR to be ≥5 A PS4 calculator is available at [www.cardiodb.org](https://www.cardiodb.org/ps4_calculator/ps4_calculator.html). If multiple cohorts are available, the final ORs and associated CIs need to be harmonized across all cohorts to determine the final level (e.g., if 2 large cohorts have an OR of ~6 and a third small cohort has an OR of 11, application at a SUPPORTING level should be considered). *RELEVANT PHENOTYPES: 1. Cases of HCM and RCM may be combined as they are considered part of the same disease spectrum. 2. For the eight genes covered by these guidelines, the combination of probands with other phenotypes should be reviewed by a clinical expert to determine if grouping is appropriate. 3. Additional considerations for LVNC and end-stage HCM: * Due to the current debate about whether isolated LVNC represents a true disease entity or variation of typical cardiac morphology (Anderson et al. 2017[<sup>9</sup>](#pmid_28395867); Oechslin et al. 2017[<sup>6</sup>](#pmid_28545618); Hershberger et al. 2017[<sup>5</sup>](#pmid_29212902); Ross et al. 2020[<sup>7</sup>](#pmid_31143950)), individuals with isolated LVNC should NOT be added to proband or segregation counts (including individuals with isolated LVNC in a family with other cardiomyopathies). HCM and DCM have distinct mechanisms of disease and therefore pathogenetic variants are not anticipated to cause both primary phenotypes. While occurrence in both phenotypes may initially be considered as evidence against pathogenicity, end-stage HCM can present similarly to DCM. Careful consideration is needed before including DCM or related phenotypes in case or segregation data for primarily HCM variants.	Curator only: Needs case–control data or counts of affected probands.

PTEN: PTEN VCEP, version 3.2 (GN003)

ClinGen PTEN Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PTEN Version 3.2. Released, last updated 2026-04-06; disease PTEN hamartoma tumor syndrome (MONDO:0017623, autosomal dominant inheritance). Residues and exons are read on NM_000314.8 (ENST00000371953.8). Specification GN003(opens in a new tab)

28 rows

ClinGen PTEN Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PTEN Version 3.2: criteria and how the engine applies them
Code	In the engine	PTEN VCEP rule (v3.2)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Use PTEN PVS1 decision tree. · Strong: Use PTEN PVS1 decision tree. · Moderate: Use PTEN PVS1 decision tree.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change OR different variant at same nucleotide position as a pathogenic splicing variant, where in silico models predict impact equal to or greater than the known pathogenic variant.	PS1 Strong only.
`PS2`	Curator only	Very Strong: Two proven OR four assumed OR one proven + two assumed de novo observations in a patient with the disease and no family history. · Strong: De novo (both maternity and paternity confirmed) observation in a patient with the disease and no family history.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product. * RNA, mini-gene, or other assay shows impact on splicing · Moderate: Well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product. * Phosphatase activity ≤ -1.11 per Mighell et al. 2018, PMID: 29706350. · Supporting: Phosphatase activity <50% of wild-type or abnormal in vitro cellular assay or transgenic model with phenotype different from wild type that does not meet PS3_moderate.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Very Strong: Probands with specificity score ≥16 (see text). · Strong: Probands with specificity score 4-15.5 (see text) OR The prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls. · Moderate: Probands with specificity score of 2-3.5 (see text). · Supporting: Phenotype specific for disease with single genetic etiology. Proband(s) with specificity score of 1-1.5 (see text).	Curator only: Needs case–control data or counts of affected probands.

CDH23: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease Usher syndrome (MONDO:0019501, autosomal recessive inheritance); nonsyndromic genetic hearing loss (MONDO:0019497, autosomal recessive inheritance). Residues and exons are read on NM_022124.6 (ENST00000224721.12). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

COCH: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease nonsyndromic genetic hearing loss (MONDO:0019497, autosomal dominant inheritance). Residues and exons are read on NM_004086.3 (ENST00000396618.9). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

GJB2: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease nonsyndromic genetic hearing loss (MONDO:0019497, autosomal recessive inheritance). Residues and exons are read on NM_004004.6 (ENST00000382848.5). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

KCNQ4: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease nonsyndromic genetic hearing loss (MONDO:0019497, autosomal dominant inheritance). Residues and exons are read on NM_004700.4 (ENST00000347132.10). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

MYO7A: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease Usher syndrome (MONDO:0019501, autosomal recessive inheritance); nonsyndromic genetic hearing loss (MONDO:0019497, autosomal recessive inheritance). Residues and exons are read on NM_000260.4 (ENST00000409709.9). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

SLC26A4: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease Pendred syndrome (MONDO:0010134, autosomal recessive inheritance). Residues and exons are read on NM_000441.2 (ENST00000644269.2). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

USH2A: Hearing Loss VCEP, version 2.0 (GN005)

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0. Released, last updated 2022-03-30; disease Usher syndrome (MONDO:0019501, autosomal recessive inheritance). Residues and exons are read on NM_206933.4 (ENST00000307340.8). Specification GN005(opens in a new tab)

28 rows

ClinGen Hearing Loss Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for CDH23, COCH, GJB2, KCNQ4, MYO6, MYO7A, SLC26A4, TECTA and USH2A Version 2.0: criteria and how the engine applies them
Code	In the engine	Hearing Loss VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant in a gene with established LOF as a disease mechanism; see PVS1_Strong, PVS1_Moderate, PVS1_Supporting for reduced evidence applications. · Strong: See PVS1 flow chart for PVS1_Strong variants in gene where LOF is a known mechanism of disease. * PVS1 should also be considered for the following genes with variants assessed in the Hearing Loss Variant Pilot: GJB2, CDH23, USH2A, SLC26A4, MYO6, MYO7A, TECTA, KCNQ4. * For other genes, LOF must be an established disease mechanism, and the gene/disease association must be Strong or Definitive clinical validity level as outlined in Strande et al. 2017 (PMID: 28552198). * If above criteria is met, follow PVS1 flowchart as recommended by the SVI. · Moderate: See PVS1 flowchart for PVS1_Moderate variants in gene where LOF is a known mechanism of disease. · Supporting: See PVS1 flowchart for PVS1_Supporting variants in gene where LOF is a known mechanism of disease.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as an established pathogenic variant; OR splice variants at same nucleotide and with similar impact prediction as previously reported pathogenic variant. * Established variant must meet criteria for pathogenicity by the HL specifications * Can also use PS1 for splice variants located in the splice consensus sequence, at the same nucleotide position as a previously reported pathogenic variant * Example: c.105+1G>C is known to be pathogenic, can use PS1 for c.105+1G>T * No additional hearing loss specifications for missense variants. Follow recommendations as outlined in Richard 2015 and/or the Sequence Variant Interpretation working group within ClinGen. * Caveat (from ACMG/AMP guidelines): Assess the possibility that the variant may act directly through the DNA change (e.g. through splicing disruption as assessed by at least computational analysis) instead of through the amino acid change)	PS1 Strong only.
`PS2`	Curator only	Very Strong: 4 points per tables 5a and 5b: Examples: 2 proven de novo occurrences; OR 1 proven + 2 assumed de novo occurrences; OR 4 assumed de novo occurrences. · Strong: 2 points per tables 5a and 5b: Examples: 1 proven de novo occurrence; OR 2 assumed de novo occurrences. · Moderate: 1 point per tables 5a and 5b: Examples: 1 proven de novo occurrence (phenotype consistent but not specific to gene); OR 1 assumed de novo occurrence; OR 2 assumed de novo occurrences (phenotype/gene not specific). · Supporting: 0.5 points per tables 5a and 5b: Example: 1 assumed de novo occurrence (phenotype/gene not specific).	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Knock-in mouse model demonstrates the phenotype. · Moderate: Validated functional studies show a deleterious effect (predefined list): GJB2: electrical coupling assays, dye transfer assays → PS3_Moderate * Dye Transfer Assays: Expect results that compare the fluorescence of a variant-transfected cell to both a negative control (or H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in no dye transfer or significantly different dye transfer when compared to the wildtype. * Electrical Coupling Assays: Expect results comparing the current of the variant-transfected cells to both a negative control (i.e. H2O injected control) and a wildtype-transfected cell. PS3_Moderate would be applied if the variant results in significantly different current compared to the wildtype, and the current is comparable to background levels/negative control. · Supporting: SLC26A4: Radio isotope and fluorescence assays → PS3_Supporting * Radio Isotope Assays: PS3_Supporting would be applied when cells transfected with mutant SLC26A4 show a statistically significant decreased efflux of iodide compared to wildtype pendrin. * Fluorescence Assays: PS3_Supporting would be applied when a cell transfected with the mutant SLC26A4 shows a statistically significant difference in fluorescence (ΔFmax %) compared to the wildtype protein, and when the fluorescence is not significantly different from that of an empty vector control. COCH: Localization, secretion, and dimerization studies performed using immunofluorescence and Western blotting techniques →PS3_Supporting * Localization: PS3_Supporting would be applied if the mutant cochlin protein does not aggregate into extracellular deposits or in the perinuclear region, comparable to the localization of wildtype cochlin. * Secretion: PS3_Supporting would be applied if cochlin protein containing the variant does not show secretion from transfected cells, but aggregates in cell regions such as the ER, Golgi and nucleus or is degraded. * Dimerization: In a non-reducing environment, wildtype cochlin migrate quickly and appear smaller than in the reduced state because the structure is maintained by disulfide bonds. PS3_Supporting would be applied if the cochlin protein containing the variant forms more, or less, stable disulfide bonds when compared to the wildtype in non-reducing conditions. * If not listed above, OK to use PS3_Supporting for other genes/functional analyses if * The assay has been validated by a known pathogenic and benign variant AND * There is plausible reason that the function the assay is testing relates to the phenotype AND * The assay conditions are likely to mimic the physiological environment.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: Fisher Exact or Chi-Squared analysis shows statistical increase in cases over controls, OR Autosomal dominant: ≥15 probands with variant, and variant meets PM2_Supporting. · Moderate: Autosomal dominant: ≥6 probands with variant, and variant meets PM2_Supporting. · Supporting: Autosomal dominant: ≥2 probands with variant, and variant meets PM2_Supporting.	Curator only: Needs case–control data or counts of affected probands.

PAH: Phenylketonuria VCEP, version 2.0 (GN006)

ClinGen Phenylketonuria Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PAH Version 2.0. Released, last updated 2024-07-16; disease phenylketonuria (MONDO:0009861, ). Residues and exons are read on NM_000277.3 (ENST00000553106.6). Specification GN006(opens in a new tab)

28 rows

ClinGen Phenylketonuria Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PAH Version 2.0: criteria and how the engine applies them
Code	In the engine	Phenylketonuria VCEP rule (v2.0)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Applicable as described in Tayoun et al. 2018. * Any nonsense or frameshift variant occurring upstream of c.1285 * Any canonical splice site predicted to disrupt reading frame and undergo nonsense mediated decay PVS1 (RNA): splicing assay data - assays demonstrating a variant leads to aberrant splicing profile that can be categorized against a PVS1 decision tree * Use the PVS1 decision tree to determine code strength * Applicable as described in Walker et al. (PMID: 36865205) · Strong: Use PVS1_strong with: * Any nonsense or frameshift variant occurring downstream of c.1285 * Any canonical splice site predicted to preserve reading frame (skipping of exons 1, 9, 10) or affect the last exon (exon 13) * Initiator codon variant	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Same predicted splicing impact as a previously classified (likely) pathogenic variant Applicable as described in Walker et al. (PMID: 36865205)	PS1 Strong; the panel adds splicing comparisons (curator).
`PS2`	Curator only	Strong: Confirmation of paternity only is insufficient. Egg donation, surrogate motherhood, errors in embryo transfer, etc. can contribute to non-maternity. Only applicable when proband has a known pathogenic variant in trans with the de novo variant.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Functional studies with sufficient analyses to calculate OddsPath reaching strong have not been identified. Therefore, the strength of this criteria is modified to PS3_moderate or PS3_supporting for future or existing studies. In vitro enzyme activity <50% compared to wild type controls. * Expression systems placing the mutant (and wild-type) cDNAs into plasmid vectors and introducing these into human or other mammalian host cells, which is the closest available approximation to the in vivo situation (e.g., COS cells) (Trunzo et al. Gene. 2016. 594:138-143. PMID: 27620137). * With ≥11 benign/pathogenic variant controls used in assay * NOTE: no papers that meet PS3_Moderate criteria have been identified by the PAH VCEP at time of this specification update. However, there may be future studies that meet the above criteria where a moderate level of evidence can be applied. · Supporting: In vitro enzyme activity ≤50% compared to wild type controls * with ≤10 benign/pathogenic variant controls used in assay	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Not applicable	Not applicable for PAH (Phenylketonuria VCEP v2.0).	Not applicable for this gene: shown, never scored.

CDH1: Gastric Cancer VCEP, version 3.1 (GN007)

ClinGen CDH1 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines Version 3.1. Released, last updated 2022-03-29; disease hereditary diffuse gastric adenocarcinoma (MONDO:0007648, ). Residues and exons are read on NM_004360.5 (ENST00000261769.10). Specification GN007(opens in a new tab)

28 rows

ClinGen CDH1 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines Version 3.1: criteria and how the engine applies them
Code	In the engine	Gastric Cancer VCEP rule (v3.1)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Per modified CDH1 PVS1 decision tree. · Strong: Per modified CDH1 PVS1 decision tree. Other CDH1 caveats: * Use PVS1_Strong as the default strength of evidence for canonical splice site variants and follow the site-specific recommendations in the splicing table. * CDH1 Exonic deletions or tandem duplications of in-frame exons (exon 4,5,8,9,12,13,15). · Moderate: Per modified CDH1 PVS1 decision tree. Other CDH1 caveats: * G to non-G variants disrupting the last nucleotide of an exon. * Canonical splice sites predicted or demonstrated experimentally to result in in-frame partial skipping/insertion (e.g., Exon 3 donor site).	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Not applicable	Not applicable for CDH1 (Gastric Cancer VCEP v3.1).	Not applicable for this gene: shown, never scored.
`PS2`	Curator only	Very Strong: ≥Two patients meet the HDGC individual phenotype criteria w/ parental confirmation. · Strong: One patient meets the HDGC individual phenotype criteria w/ parental confirmation.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: RNA assay demonstrating abnormal out-of-frame transcripts. · Moderate: RNA assay demonstrating abnormal in-frame transcript.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Very Strong: ≥Sixteen families meet HDGC criteria. · Strong: Four - Fifteen families meet HDGC criteria. · Moderate: Two or three families meet HDGC criteria. · Supporting: One family meets HDGC criteria.	Curator only: Needs case–control data or counts of affected probands.

RUNX1: Myeloid Malignancy VCEP, version 3.1 (GN008)

ClinGen Myeloid Malignancy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RUNX1 Version 3.1. Released, last updated 2026-02-13; disease hereditary thrombocytopenia and hematologic cancer predisposition syndrome (MONDO:0011071, autosomal dominant inheritance). Residues and exons are read on NM_001754.5 (ENST00000675419.1). Specification GN008(opens in a new tab)

28 rows

ClinGen Myeloid Malignancy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RUNX1 Version 3.1: criteria and how the engine applies them
Code	In the engine	Myeloid Malignancy VCEP rule (v3.1)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Per modified RUNX1 PVS1 decision tree for SNVs and CNVs and table of splicing effects. · Strong: Per modified RUNX1 PVS1 decision tree for SNVs and CNVs and table of splicing effects. · Moderate: Per modified RUNX1 PVS1 decision tree for SNVs and CNVs and table of splicing effects.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: PS1: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. For splice site variants: PS1_Variable: Follow recommendations from the ClinGen SVI Splicing Subgroup (Walker et al., 2023, PMID: 37352859 [<sup>5</sup>](#PMID_37352859)) · Moderate: PS1_Moderate: Same amino acid change as a previously established likely pathogenic variant regardless of nucleotide change. For splice site variants: PS1_Variable: Follow recommendations from the ClinGen SVI Splicing Subgroup (Walker et al., 2023, PMID: 37352859 [<sup>5</sup>](#PMID_37352859))	PS1 Strong (Pathogenic) or Moderate (Likely pathogenic).
`PS2`	Curator only	Moderate: PS2_Moderate: ≥ 2 proven de novo occurrences (both maternity and paternity confirmed) in patients with FPD/AML phenotype. · Supporting: PS2_Supporting: 1 proven de novo occurrence (both maternity and paternity confirmed) in a patient with FPD/AML phenotype.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: PS3: Transactivation assays demonstrating altered transactivation (<20% of wt, and/or reduced to levels similar to well established pathogenic variants such as R201Q or R166Q) AND data from a secondary assay demonstrating altered function. Not applicable if variant meets PVS1. If variant meets PVS1_strong, upgrade to PVS1. · Moderate: PS3_Moderate: Transactivation assays demonstrating altered transactivation (<20% of wt and/or reduced to levels similar to well established pathogenic variants such as R201Q or R166Q) OR ≥ 2 secondary assays demonstrating altered function. · Supporting: PS3_Supporting: Transactivation assays demonstrating enhanced transactivation (>115% of wt).	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: PS4: ≥ 4 probands meeting at least one of the RUNX1-phenotypic criteria (OR 127.1). · Moderate: PS4_Moderate: 2-3 probands meeting at least one of the RUNX1-phenotypic criteria (OR 63.5-95.3). · Supporting: PS4_Supporting: 1 proband meeting at least one of the RUNX1-phenotypic criteria (OR 31.8).	Curator only: Needs case–control data or counts of affected probands.

TP53: TP53 VCEP, version 2.4 (GN009)

ClinGen TP53 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for TP53 Version 2.4. Released, last updated 2025-11-20; disease Li-Fraumeni syndrome (MONDO:0018875, autosomal dominant inheritance). Residues and exons are read on NM_000546.6 (ENST00000269305.9). Specification GN009(opens in a new tab)

28 rows

ClinGen TP53 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for TP53 Version 2.4: criteria and how the engine applies them
Code	In the engine	TP53 VCEP rule (v2.4)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Please utilize the PVS1 decision tree for application of PVS1 code. The decision tree details the specific strengths each type of null variant may be applied at. Please see below for some additional helpful summary details for application of PVS1 code: * Initiation codon: * PVS1 may be applied to initiation codon variants * Nonsense or frameshift variants: * PVS1 applies to variants predicted to result in nonsense-mediated decay (NMD) for nonsense variants upstream of p.Lys351 and for frameshift induced premature termination codon (PTC) upstream of p.Lys351 * PVS1_Strong applies to variants not predicted to undergo NMD (nonsense variants downstream of codon 350 or frameshift induced PTC in exon 11 or in the 3’ most 50 nucleotides of exon 10) in variants located in the p.Lys351 to p.Ala 355 range * PVS1_Moderate applies to variants not predicted to undergo NMD (nonsense variants downstream of codon 350 or frameshift induced PTC in exon 11 or in the 3’ most 50 nucleotides of exon 10) in variants located in the p.Gly356 to p.Asp393 range * PVS1_Moderate may also be applied to frameshift induced PTC downstream of the natural stop codon * Canonical splice variants (+/- 1,2 intronic positions): * PVS1 applies to predicted splicing alterations that are PTC resulting in NMD (or in-frame but targeting critical domains or residues) * PVS1 applies to predicted splicing alterations that target the start codon (Exon 2 donor) * PVS1_Moderate applied to splicing alterations that are predicted to shorten (<10% of the protein removed) or expand a TP53 C-terminal end of unknown function (E10 donor or E11 acceptor) * Deletions * Full gene deletions: PVS1 * Single- to multi-exon deletions that target the initiation codon, preserving the potential rescue ATG (p.Met40) in exon 4: PVS1 * Single- to multi-exon deletions that target the initiation codon and the potential rescue ATG (p.Met40) in exon 4: PVS1 * Single- to multi-exon deletion that disrupts the reading frame and is predicted to undergo NMD (nonsense or frameshift induced PTC upstream of p.Lys351): PVS1 * Single- to multi-exon deletion that disrupts the reading frame and is NOT predicted to undergo NMD (nonsense or frameshift inducted PTC downstream of p.Leu350): PVS1 * Single- to multi-exon deletion including the last exon where the truncated/altered region is critical to protein function (any multi-exon combination targeting exon 11): PVS1 * If the role of the region in protein function is unknown, if the variant removed < 10% of the protein (deletion of exon 11): PVS1_Moderate * Single- to multi-exon deletion that preserves the reading frame where the truncated/altered region is critical to protein function: PVS1 * Duplications (≥1 exon in size and must be completely contained within the TP53 gene) * Proven in tandem. Reading frame is disrupted and NMD predicted to occur (nonsense upstream of p.Lys351 or frameshift-induced PTC upstream of p.Lys351): PVS1 * Presumed in tandem. Reading frame presumed disrupted and NMD predicted to occur (nonsense upstream of p.Lys351 or frameshift-induced PTC upstream of p.Lys351): PVS1_Strong For variants inducing aberrant transcripts identified via mRNA assay, apply as PVS1_Variable Weight (RNA) following recommendations from Walker et al., 2023 (PMID: 37352859), downgrading one strength level if the assay data indicates leakiness. Caveats: PS3 should not be applied at any strength if PVS1 is applied at full strength. PP3 should not be used in combination with PVS1. For the purposes of unified curation, the TP53 domains/important motifs by amino acid range are defined as: TAD1: aa 17-25 TAD2: aa 48-56 Proline residues: aa 64-92 DNA binding domain: aa 100-292 Hinge domain: aa 293-324 Oligomerization domain: aa 325-356 C-terminal domain (Basic domain): aa 368-387 A disease-specific PVS1 decision tree incorporating the above bullets as well as a supplemental file for TP53 PVS1 Splicing Worksheet is also included as an additional curation tool and has more granular details. · Strong: See PVS1 flowchart for code application · Moderate: See PVS1 flowchart for code application	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Can be applied to variants asserted as Pathogenic following the TP53 VCEP’s specifications. · Moderate: Can be applied to variants asserted as Likely Pathogenic following the TP53 VCEP’s specifications.	PS1 Strong (Pathogenic) or Moderate (Likely pathogenic).
`PS2`	Curator only	Very Strong: ≥ 8 points · Strong: 4-7 points · Moderate: 2-3 points · Supporting: 1 point	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Non-functional on Kato et al. data AND loss of function (LOF) by the majority of other eligible assays · Moderate: Partially functional on Kato et al. data AND loss of function (LOF) by the majority of other available assays · Supporting: Non-functional on Kato et al. data AND abnormal on Kawaguchi et al. data regardless of other assays PS3_Supporting may also be applied to small deletions that demonstrate LOF on the majority of available assays	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Very Strong: ≥ 8 points · Strong: ≥ 4-7.5 points · Moderate: 2-3.5 points · Supporting: 1-1.5 points	Curator only: Needs case–control data or counts of affected probands.

MYOC: Glaucoma VCEP, version 2.1 (GN019)

ClinGen Glaucoma Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MYOC Version 2.1. Released, last updated 2025-11-06; disease open-angle glaucoma (MONDO:0005338, autosomal dominant inheritance). Residues and exons are read on NM_000261.2 (ENST00000037502.11). Specification GN019(opens in a new tab)

28 rows

ClinGen Glaucoma Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MYOC Version 2.1: criteria and how the engine applies them
Code	In the engine	Glaucoma VCEP rule (v2.1)	How the engine applies it
`PVS1`	Not applicable	Not applicable for MYOC (Glaucoma VCEP v2.1).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as previously established pathogenic variant · Moderate: Same amino acid change as a previously established likely pathogenic variant	PS1 Strong (Pathogenic) or Moderate (Likely pathogenic).
`PS2`	Curator only	Strong: * ≥2 confirmed de novo in JOAG · Moderate: * ≥2 confirmed de novo in POAG * Or 1 confirmed de novo in JOAG * Or ≥2 assumed de novo in JOAG · Supporting: * 1 confirmed de novo in POAG * Or ≥2 assumed de novo in POAG * Or 1 assumed de novo in JOAG	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Strong: Assays with OddsPath >18.7 as per the SVI recommendations.[<sup>1</sup>](#PMID_31892348) · Moderate: Assays with OddsPath >4.3 as per the SVI recommendations.[<sup>1</sup>](#PMID_31892348) · Supporting: Assays with OddsPath >2.1 as per the SVI recommendations.[<sup>1</sup>](#PMID_31892348)	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥ 15 probands from multiple independent studies. · Moderate: ≥ 6 probands from multiple independent studies. · Supporting: ≥ 2 probands from multiple independent studies.	Curator only: Needs case–control data or counts of affected probands.

SHOC2: RASopathy VCEP, version 2.3 (GN038)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SHOC2 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_007373.4 (ENST00000369452.9). Specification GN038(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SHOC2 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for SHOC2 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant in SHOC2 regardless of nucleotide change.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Not applicable	Not applicable for SHOC2 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

NRAS: RASopathy VCEP, version 2.3 (GN039)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for NRAS Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_002524.5 (ENST00000369535.5). Specification GN039(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for NRAS Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for NRAS (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous pathogenic residue positions in HRAS, KRAS, MRAS, NRAS, RIT1, and RRAS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

RAF1: RASopathy VCEP, version 2.3 (GN040)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RAF1 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_002880.4 (ENST00000251849.9). Specification GN040(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RAF1 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for RAF1 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in BRAF and RAF1.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

SOS1: RASopathy VCEP, version 2.3 (GN041)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SOS1 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_005633.4 (ENST00000402219.8). Specification GN041(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SOS1 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for SOS1 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in SOS1 and SOS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

SOS2: RASopathy VCEP, version 2.3 (GN042)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SOS2 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_006939.4 (ENST00000216373.10). Specification GN042(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for SOS2 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for SOS2 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in SOS1 and SOS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

PTPN11: RASopathy VCEP, version 2.3 (GN043)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PTPN11 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_002834.5 (ENST00000351677.7). Specification GN043(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for PTPN11 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for PTPN11 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant in PTPN11 regardless of nucleotide change.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

KRAS: RASopathy VCEP, version 2.3 (GN044)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for KRAS Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_004985.5 (ENST00000311936.8). Specification GN044(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for KRAS Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for KRAS (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in HRAS, KRAS, MRAS, NRAS, RIT1, and RRAS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

MAP2K1: RASopathy VCEP, version 2.3 (GN045)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MAP2K1 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_002755.4 (ENST00000307102.10). Specification GN045(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MAP2K1 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for MAP2K1 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in MAP2K1 and MAP2K2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

HRAS: RASopathy VCEP, version 2.3 (GN046)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for HRAS Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_005343.4 (ENST00000311189.8). Specification GN046(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for HRAS Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for HRAS (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in HRAS, KRAS, MRAS, NRAS, RIT1, and RRAS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

RIT1: RASopathy VCEP, version 2.3 (GN047)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RIT1 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_006912.6 (ENST00000368323.8). Specification GN047(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for RIT1 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for RIT1 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in HRAS, KRAS, MRAS, NRAS, RIT1, and RRAS2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 point.	Curator only: Needs case–control data or counts of affected probands.

MAP2K2: RASopathy VCEP, version 2.3 (GN048)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MAP2K2 Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_030662.4 (ENST00000262948.10). Specification GN048(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for MAP2K2 Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for MAP2K2 (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in MAP2K1 and MAP2K2.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous residues in related genes are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

BRAF: RASopathy VCEP, version 2.3 (GN049)

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRAF Version 2.3. Released, last updated 2024-12-03; disease RASopathy (MONDO:0021060, autosomal dominant inheritance). Residues and exons are read on NM_004333.6 (ENST00000646891.2). Specification GN049(opens in a new tab) · RASopathy expert panel methods (Gelb et al. 2018)(opens in a new tab)

28 rows

ClinGen RASopathy Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRAF Version 2.3: criteria and how the engine applies them
Code	In the engine	RASopathy VCEP rule (v2.3)	How the engine applies it
`PVS1`	Not applicable	Not applicable for BRAF (RASopathy VCEP v2.3).	Not applicable for this gene: shown, never scored.
`PS1`	Automated (gene-specific)	Strong: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change. Applicable for observed analogous residue positions in BRAF and RAF1.	PS1 Strong only: same amino-acid change as an established pathogenic variant; analogous BRAF/RAF1 residues are left to a curator.
`PS2`	Curator only	Very Strong: 4 Points. · Strong: 2 Points. · Moderate: 1 Point.	Curator only: De novo status needs parental testing.
`PS3`	Curator only	Moderate: Two or more different approved assays. · Supporting: One approved assay.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: ≥5 points. · Moderate: ≥3 points. · Supporting: ≥1 points.	Curator only: Needs case–control data or counts of affected probands.

BRCA1: ENIGMA BRCA1 and BRCA2 VCEP, version 1.2 (GN092)

ClinGen ENIGMA BRCA1 and BRCA2 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRCA1 Version 1.2. Released, last updated 2025-01-09; disease BRCA1-related cancer predisposition (MONDO:0700268, autosomal dominant inheritance). Residues and exons are read on NM_007294.4 (ENST00000357654.9). Specification GN092(opens in a new tab)

28 rows

ClinGen ENIGMA BRCA1 and BRCA2 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRCA1 Version 1.2: criteria and how the engine applies them
Code	In the engine	ENIGMA BRCA1 and BRCA2 VCEP rule (v1.2)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Strong: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Moderate: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Supporting: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Apply PS1, for predicted missense substitutions, where a previously classified pathogenic variant is considered to act via protein change (no confirmed or predicted effect on mRNA splicing (SpliceAI≤0.1)). Apply PS1, for exonic and intronic variants with same predicted impact on splicing, as a previously classified pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details. · Moderate: Apply PS1_Moderate, for predicted missense substitutions, where previously classified likely pathogenic variant is considered to act via protein change (no confirmed or predicted effect on mRNA splicing (SpliceAI≤0.1)). Apply PS1_Moderate, for exonic and intronic variants with same predicted impact on splicing, as a previously classified (likely) pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details. · Supporting: Apply PS1_Supporting, for exonic and intronic variants with same predicted impact on splicing, as a previously classified (likely) pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details.	PS1 Strong (Pathogenic) or Moderate (Likely pathogenic) for missense acting through protein change.
`PS2`	Not applicable	Not applicable for BRCA1 (ENIGMA BRCA1 and BRCA2 VCEP v1.2).	Not applicable for this gene: shown, never scored.
`PS3`	Curator only	Strong: Well-established in vitro or in vivo functional studies supportive of a damaging effect. Apply PS3 for assays measuring effect via protein only OR mRNA and protein combined. See Specifications Table 9 for code recommendations from calibrated published assays. Also see Figure1C and Appendix E for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: The prevalence of the variant in affected individuals is significantly increased compared to the prevalence in controls. Case-control studies; p-value ≤0.05 and OR ≥4 (lower confidence interval excludes 2.0). See Appendix F for details.	Curator only: Needs case–control data or counts of affected probands.

BRCA2: ENIGMA BRCA1 and BRCA2 VCEP, version 1.2 (GN097)

ClinGen ENIGMA BRCA1 and BRCA2 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRCA2 Version 1.2. Released, last updated 2025-01-09; disease BRCA2-related cancer predisposition (MONDO:0700269, autosomal dominant inheritance). Residues and exons are read on NM_000059.4 (ENST00000380152.8). Specification GN097(opens in a new tab)

28 rows

ClinGen ENIGMA BRCA1 and BRCA2 Expert Panel Specifications to the ACMG/AMP Variant Interpretation Guidelines for BRCA2 Version 1.2: criteria and how the engine applies them
Code	In the engine	ENIGMA BRCA1 and BRCA2 VCEP rule (v1.2)	How the engine applies it
`PVS1`	Automated (generic)	Very Strong: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Strong: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Moderate: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details. · Supporting: Null variant (nonsense, frameshift, splice site (donor/acceptor +/−1,2), initiation codon, single or multi-exon deletion) in a gene where loss of function (LOF) is a known mechanism of disease. Apply at appropriate strength according to PVS1 flowchart, which considers knowledge of clinically important functional domains. See Specifications Table 4 and Appendix D for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details.	Automated: Stop-gain, frameshift, canonical splice or start-loss in a gene with ClinGen haploinsufficiency 3 or 30, graded by a simplified ClinGen PVS1 tree on LOFTEE: Very Strong (high confidence, NMD predicted), Strong (NMD escape, or LOFTEE flags), Moderate (start-loss, last 10% of the protein, or END_TRUNC). Anything else goes to a curator.
`PS1`	Automated (gene-specific)	Strong: Apply PS1, for predicted missense substitutions, where a previously classified pathogenic variant is considered to act via protein change (no confirmed or predicted effect on mRNA splicing (SpliceAI≤0.1)). Apply PS1, for exonic and intronic variants with same predicted impact on splicing, as a previously classified pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details. · Moderate: Apply PS1_Moderate, for predicted missense substitutions, where a previously classified likely pathogenic variant is considered to act via protein change (no confirmed or predicted effect on mRNA splicing (SpliceAI≤0.1)). Apply PS1_Moderate, for exonic and intronic variants with same predicted impact on splicing, as a previously classified (likely) pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details. · Supporting: Apply PS1, for exonic and intronic variants with same predicted impact on splicing, as a previously classified (likely) pathogenic variant. Vary weight depending on relative positions, and confidence in classification of the reference variant. See Specifications Table 5 and Appendix E, J and K for details.	PS1 Strong (Pathogenic) or Moderate (Likely pathogenic) for missense acting through protein change.
`PS2`	Not applicable	Not applicable for BRCA2 (ENIGMA BRCA1 and BRCA2 VCEP v1.2).	Not applicable for this gene: shown, never scored.
`PS3`	Curator only	Strong: Well-established in vitro or in vivo functional studies supportive of a damaging effect. Apply PS3 for assays measuring effect via protein only OR mRNA and protein combined. See Specifications Table 9 for code recommendations from calibrated published assays. Also see Figure1C and Appendix E for details. Well-established in vitro or in vivo functional studies supportive of a damaging effect as measured by effect on mRNA transcript profile (mRNA assay only). Apply as PVS1 (RNA) at appropriate strength. See Specifications Figure1B and Appendix E for details.	Curator only: Functional studies are curator-only (Brnich et al. 2020).
`PS4`	Curator only	Strong: The prevalence of the variant in affected individuals is significantly increased compared to the prevalence in controls. Case-control studies; p-value ≤0.05 and OR ≥4 (lower confidence interval excludes 2.0). See Appendix F for details.	Curator only: Needs case–control data or counts of affected probands.

All ClinGen specifications by gene

226 rows

ClinGen criteria specifications by gene (CSpec Registry)
Gene	Specification	Expert panel	Version	Status	Disease (MONDO)	Inheritance	Applied by the engine
`ABCA4`	GN164(opens in a new tab)	ABCA4 VCEP	1.0	Released	MONDO:0800406	AR	No
`ABCD1`	GN105(opens in a new tab)	Peroxisomal Disorders VCEP	1.0	Released	MONDO:0018544	XL	No
`ACADVL`	GN021(opens in a new tab)	ACADVL VCEP	2.2	Released	MONDO:0008723	AR	No
`ACO2`	GN211(opens in a new tab)	Optic Nerve Atrophy VCEP	1.0	Classification Rules In Prep	MONDO:0014571	SD	No
`ACTA1`	GN147(opens in a new tab)	Congenital Myopathies VCEP	2.0	Released	MONDO:0100084	AD	No

ClinGen expert-panel classifications

For every variant the engine looks up its ClinGen Allele Registry identifier (CAid) and asks the ClinGen Evidence Repository for classifications published by ClinGen expert panels for exactly that allele. When there is one, the variant page shows the panel, the condition, the classification, the criteria the panel applied, the date and a link to the full record, above the automated result. An expert-panel classification is a curated, peer-reviewed assertion and takes precedence; the automated result is shown beside it so the difference can be read (usually the case-level, de novo and functional evidence that only curators can apply).

What is never automated

Case-level evidence: PS4, PM1, PP2, BP1, BP3, BP7, PS2, PM6, PS3, BS3, PM3, BP2, PP1, BS4, PP4, BP5. These need patients, families, phenotypes, phase or laboratory studies.
BS1 and BS2 are raised for review but never scored without a disease-specific threshold: a frequency above 1% is high for most rare disorders but not all, and gnomAD individuals are not confirmed unaffected.
Splicing impact: no licence-free splice predictor is used, so BP7 and splice-based PP3/BP4 are not evaluated.
Repeat regions (BP3) and analogous residues in paralogous genes (for PS1) are not annotated automatically.
Mitochondrial variants, alleles longer than 50 bases, structural variants and copy-number changes are not classified by this engine.
Somatic interpretation is separate: this page covers germline classification only.

Data sources

gnomAD v4.1(opens in a new tab)
Allele frequencies, grpmax FAF, homozygotes, coverage. Terms: ODbL 1.0.
MANE v1.5(opens in a new tab)
MANE Select and MANE Plus Clinical transcripts (variant naming, gene-specific residues). Terms: Public domain (NCBI) / Ensembl: no restrictions.
Ensembl VEP(opens in a new tab)
Consequence and LOFTEE when gnomAD has no annotation on the MANE Select transcript. Terms: Ensembl data: no restrictions.
dbNSFP 4.8a via MyVariant.info(opens in a new tab)
AlphaMissense and BayesDel_noAF scores. Terms: AlphaMissense CC BY 4.0; dbNSFP academic build (check before commercial use).
ClinVar(opens in a new tab)
Classified variants at the same residue; the variant’s own classification (comparison only). Terms: Public domain (NCBI).
ClinGen Dosage Sensitivity(opens in a new tab)
Haploinsufficiency scores (loss-of-function mechanism). Terms: Free public data.
GenCC(opens in a new tab)
Gene–disease validity. Terms: CC0.
ClinGen CSpec Registry(opens in a new tab)
Gene-specific ACMG/AMP rules of ClinGen expert panels. Terms: CC0 1.0 (attribution requested).
ClinGen Evidence Repository(opens in a new tab)
Expert-panel classifications, via the ClinGen Allele Registry. Terms: CC0 1.0 (attribution requested).

References

Richards S, et al. Standards and guidelines for the interpretation of sequence variants (ACMG/AMP). Genet Med 2015;17:405–424. doi:10.1038/gim.2015.30(opens in a new tab) · PMID 25741868
Tavtigian SV, et al. Fitting a naturally scaled point system to the ACMG/AMP variant classification guidelines. Hum Mutat 2020;41:1734–1737. doi:10.1002/humu.24088(opens in a new tab) · PMID 32720330
Ghosh R, et al. Updated recommendation for the benign stand-alone ACMG/AMP criterion (BA1). Hum Mutat 2018;39:1525–1530. doi:10.1002/humu.23642(opens in a new tab) · PMID 30311383
Abou Tayoun AN, et al. Recommendations for interpreting the loss of function PVS1 ACMG/AMP variant criterion. Hum Mutat 2018;39:1517–1524. doi:10.1002/humu.23626(opens in a new tab) · PMID 30192042
Biesecker LG, Harrison SM. The ACMG/AMP reputable source criteria for the interpretation of sequence variants (PP5 and BP6). Genet Med 2018;20:1687–1688. doi:10.1038/gim.2018.42(opens in a new tab) · PMID 29543229
Brnich SE, et al. Recommendations for application of the functional evidence PS3/BS3 criterion. Genome Med 2019;12:3. doi:10.1186/s13073-019-0690-2(opens in a new tab) · PMID 31892348
Pejaver V, et al. Calibration of computational tools for missense variant pathogenicity classification and ClinGen recommendations for PP3/BP4 criteria. Am J Hum Genet 2022;109:2163–2177. doi:10.1016/j.ajhg.2022.10.013(opens in a new tab) · PMID 36413997
Bergquist T, et al. Calibration of additional computational tools expands ClinGen recommendation options for variant classification with PP3/BP4 criteria. Genet Med 2025;27:101402. doi:10.1016/j.gim.2025.101402(opens in a new tab) · PMID 40084623
Gelb BD, et al. ClinGen’s RASopathy Expert Panel consensus methods for variant interpretation. Genet Med 2018;20:1334–1345. doi:10.1038/gim.2018.3(opens in a new tab) · PMID 29493581
ClinGen Sequence Variant Interpretation Working Group recommendations (PM2 at Supporting, 2020; BA1 exceptions; points system). clinicalgenome.org(opens in a new tab)