Documentation / Classification / Criteria Reference
Criteria Reference
Complete reference for all 28 ACMG evidence criteria. 19 are evaluated automatically, 9 require manual curation, and 1 (PM5) is currently disabled.
Pathogenic Evidence (16 criteria)
Null variant in gene where loss-of-function is a known disease mechanism
Conditions
Impact = HIGH
Consequence: frameshift, stop_gained, splice_acceptor, or splice_donor (canonical +/-1,2 only)
Gene constraint: pLI > 0.9 OR LOEUF < 0.35 OR gene in curated AR LoF list (~150 genes) OR ClinGen HI = 3 OR gene in curated AD ClinVar LOF list (v3.13.0)
Disease association gate: gene must have established disease association (ClinVar P/LP 2+ stars, Orphanet, ClinGen HI, AR_LOF_GENES, gene_disease_mechanism, or VCEP)
Last-exon downgrade (v3.9.0): truncating variants in last exon receive PVS1_Strong (+4 pts) instead of PVS1 Very Strong (+8 pts)
MANE Select positional guard (v3.22.0, HELIX-CR-2026-055): variants outside MANE Select/Plus Clinical CDS region do not receive PVS1 or PM4. 19,354 transcripts loaded with CDS coordinates. NULL MANE = PVS1 applied (conservative fallback). Reference: Abou Tayoun 2018 PMID:30192042
Expression-aware guard (v3.23.0, HELIX-CR-2026-056): per-exon pext from gnomAD v4.1 GTEx v10 (49 tissues) modulates PVS1. max_tissue_pext >= 0.9 = Very Strong, 0.1-0.9 = PVS1_Strong[pext], < 0.1 = blocked. PM4 blocked at pext < 0.1. Reference: Cummings 2020 PMID:32461655
Exclusions
NMD-escaping transcripts (NMD_escaping_variant). Note: NMD_transcript_variant is NOT excluded (v3.11.3)
Non-canonical splice: splice_donor_5th_base_variant, splice_donor_region_variant, splice_acceptor_5th_base_variant (v3.11.4)
Gain-of-function / dominant-negative / gain-of-expression genes (v3.18.0, HELIX-REF-003): primary gof_genes_unified view (348 GoF/DN/GoE monoallelic genes from G2P + GoFCards + manual, definitive/strong/moderate). Fallback GOF_AD_GENES (29 genes including ANKRD26, CALM2, MYH3, PRKAG2, SKI, TUBB3 with PMID-documented non-LoF mechanism). Dual-mechanism genes (SCN5A, LMNA, KCNH2, MYH7) excluded from GoF view via gof_genes_exclusive (v3.20.3) and retain PVS1 eligibility
Dual-mechanism AD/AR LoF bypass (v3.27.0, HELIX-CR-2026-079): genes with documented monoallelic LoF mechanism (gene_disease_mechanism definitive/strong/moderate) OR in CLINVAR_LOF_AD_GENES (IFT140, GCM2) bypass the AR-only het LoF guard. Het LoF in dual-mechanism gene is pathogenic via AD pathway, not carrier state
Stop-retained and stop-lost variants
HLA gene family
Homozygous reference genotypes (hom_ref) from multi-allelic VCF sites
Databases: VEP (consequence, impact, exon_number), gnomAD Constraint (pLI, LOEUF), gnomAD pext v4.1 GTEx v10 (v3.23.0), MANE Select v1.4 (v3.22.0), curated AR LoF gene list, curated AD ClinVar LOF list (v3.13.0), Orphanet, ClinVar (gene-level), ClinGen (HI score, GDV), VCEP, gof_genes_unified view (348 genes), gof_genes_exclusive view (excludes 55 dual-mechanism), GOF_AD_GENES (29 genes), gene_disease_mechanism (HELIX-REF-003), CLINVAR_LOF_AD_GENES
Reading frame rescue not evaluated. Alternative transcript usage addressed via MANE Select positional guard (v3.22.0)
Tissue-specific expression assessed via gnomAD pext (v3.23.0): exons with max_tissue_pext < 0.1 block PVS1; 0.1-0.9 downgrade to PVS1_Strong
VCEP gene-specific PVS1 applicability gate available for ~50-60 genes; generic GoF/DN guard via gof_genes_unified + GOF_AD_GENES covers other non-LoF mechanism genes
Same amino acid change as an established pathogenic variant
Conditions
ClinVar: Pathogenic or Likely Pathogenic
ClinVar review stars >= 2
Databases: ClinVar
Matches exact variant, not amino acid position (PM5 matching is disabled)
De novo variant (confirmed paternity and maternity)
Requires trio sequencing data and confirmed parental relationships.
Functional studies show a deleterious effect
Requires curation of published functional assay data.
Prevalence significantly increased in affected individuals
Requires case-control study data or odds ratios.
Located in a functional domain
Conditions
Variant overlaps a curated critical functional domain from CRITICAL_PFAM_DOMAINS list (~60 domains with documented pathogenic variant clustering)
Databases: VEP (domains)
Only curated critical domains trigger PM1 (kinase catalytic, DNA-binding, ion channel pores, enzyme active sites, VCEP-documented domains)
Generic structural domains (Caveolin, coiled-coil, DUF) excluded
Absent from controls or at extremely low frequency
Conditions
Default path (global_af): gnomAD global AF < 0.0001 (0.01%) OR absent from gnomAD (NULL = never observed in ~800K individuals)
VCEP af_grpmax path (v3.21.0, HELIX-CR-2026-047): VCEP genes with pm2_frequency_field=af_grpmax use ancestry-stratified maximum subpopulation frequency. RASopathy genes use pm2_threshold=0.0 (absent from controls). Edge case fix v3.26.5 (BB-006): af_grpmax = 0.0 treated as absent (gnomAD records AC=0 variants with af_grpmax=0.0 NOT NULL)
VCEP PM2_Supporting downgrade (v3.20.0, ClinGen SVI 2023): VCEP-specified genes apply PM2 at Supporting strength (+1 pt) instead of Moderate (+2 pts) per gene-specific specification. LP1b combining rule accommodates this downgrade
Databases: gnomAD v4.1 (global_af, af_grpmax), VCEP gene-specific specifications
af_grpmax path requires VCEP coverage for the gene; non-VCEP genes use default global_af path
PM2 fundamentally limited by gnomAD population coverage and underrepresentation of certain ancestries
Detected in trans with a pathogenic variant for recessive disorders
Conditions
Compound heterozygote candidate flag = true
ClinVar partner validation guard (v3.25.1, HELIX-CR-2026-058): trans partner must be ClinVar P/LP (clinical_significance LIKE Pathogenic% OR Likely_pathogenic%, NOT Conflicting) with review_stars >= ps1_min_stars (currently 2). Aligns with Richards 2015 Table 3: PM3 explicitly requires "detected in trans with a pathogenic variant"
ClinVar-only guard (no acmg_class dependency): avoids circular reference where partner classification depends on PM3 which depends on partner classification
Exclusions
GoF / DN gene exclusion (v3.20.2, HELIX-CR-2026-036): PM3 not applied in pure GoF/DN genes. PM3 presupposes AR mechanism per Richards 2015 Table 3
Dual-mechanism bypass (v3.20.3, HELIX-CR-2026-037): genes with both GoF/DN monoallelic AND biallelic LoF mechanism (excluded from gof_genes_exclusive view) retain PM3 for the AR LoF pathway
Databases: Pipeline-internal compound het detection, ClinVar (partner clinical_significance, review_stars), gof_genes_exclusive view, GOF_AD_GENES fallback
Inferred from genotype data without formal phasing
ClinVar partner requires 2+ review stars; partners with 1-star P/LP submissions do not satisfy PM3
Protein length change in a non-repetitive region
Conditions
In-frame insertion or deletion
Located within a Pfam functional domain
Not in a repetitive or low-complexity region
Exclusions
HLA gene family
Databases: VEP (consequence, domains)
Novel missense at known pathogenic amino acid position
Pending standardized protein-level coordinate matching in ClinVar preprocessing.
Assumed de novo without confirmation
Requires family structure information not available in single-sample analysis.
Cosegregation with disease in multiple affected family members
Requires multi-generational pedigree data.
Missense in a gene with low rate of benign missense variation
Conditions
Missense variant
Gene constraint: pLI > 0.5 AND mis_z > 2.0 (missense constraint)
Databases: VEP (consequence), gnomAD Constraint (pLI, mis_z)
Computational evidence supports a deleterious effect
Conditions
Path A (Missense -- BayesDel): consequence = missense_variant required (v3.12.1, HELIX-CR-2026-009). BayesDel_noAF with ClinGen SVI calibrated thresholds (Pejaver 2022) -- PP3_Strong (>= 0.518, +4 pts), PP3_Moderate (0.290-0.517, +2 pts), PP3_Supporting (0.130-0.289, +1 pt)
Missense relevance guard v2.0 (v3.19.0, HELIX-CR-2026-033): 5 gates determine if missense is a disease mechanism for the gene. Gate 1: mis_z > 2.0 (direct constraint). Gate 2: mis_z > 1.5 AND pLI < 0.5 (AR genes). Gate 3: mis_z > 0.5 AND pLI > 0.9 (AD LoF-intolerant). Gate 4a: gene in refdb.clinvar_missense_genes (2,136 genes with ClinVar P/LP missense evidence). Gate 4b: pLI > 0.7 AND mis_z > 0.5 AND disease association (gray-zone genes). NULL fallback: pLI > 0.5 without mis_z
PP3_Strong additionally requires established disease association (v3.8.1 gate, same 7 sources as PVS1)
PM1 + PP3 double-counting guard: when PM1 applies alongside PP3_Strong, PP3 is downgraded to PP3_Moderate (combined cap at Strong equivalent per ClinGen SVI)
Path B (Splice -- PP3_splice): 3 strength levels (v3.15.1). PP3_splice_Strong (SpliceAI >= 0.8, +4 pts), PP3_splice_Moderate (0.5-0.799, +2 pts), PP3_splice (0.2-0.499, +1 pt). PP3_splice_Strong requires same disease association gate as PP3_Strong missense
Paths A and B are independent. A variant can trigger both if it has BayesDel score and splice prediction -- subject to mutual exclusion guards below
Exclusions
PP3_splice (all levels) blocked when PVS1 is triggered at any strength (Very Strong or Strong/last-exon). Prevents double-counting LoF and splice evidence per ClinGen SVI 2023 (Walker 2023)
PP3_splice missense-tolerated guard (v3.25.7, HELIX-CR-2026-065): PP3_splice (all levels) blocked for pure missense variants without VEP splice consequence when BayesDel data is available. ClinGen SVI Section 3.2: PP3 is single criterion, not applied twice
PP3 + PP3_splice mutual exclusion (v3.25.8, HELIX-CR-2026-066): for dual-consequence variants (missense + splice_region/donor/acceptor), PP3 BayesDel blocked when PP3_splice applies. SpliceAI is the more specific tool when VEP confirms splice proximity. Combined effect: exactly one PP3 path per variant
BayesDel indeterminate range (-0.180 to 0.129): no PP3 or BP4 applied
Databases: dbNSFP 4.9c (bayesdel_noaf_score), SpliceAI (max_score), VEP (consequence), gnomAD Constraint (pLI, mis_z), refdb.clinvar_missense_genes
BayesDel_noAF excludes allele frequency to avoid circular reasoning with PM2/BA1/BS1 (Pejaver 2022)
BayesDel does not reach PP3_Very_Strong per Pejaver calibration data
PM1 + PP3 cap assumes PM1 = Moderate (2 points); extensible if VCEP upgrades PM1 strength
Patient phenotype matches gene disease association
Conditions
>= 3 patient HPO terms match the gene HPO profile
OR >= 2 matches with highly specific gene (<= 5 total HPO associations)
Databases: HPO (gene-phenotype associations)
Requires patient HPO terms to be provided.
Reputable source reports variant as pathogenic
Conditions
ClinVar: Pathogenic or Likely Pathogenic
Review stars >= 1 AND < 2 (lower confidence than PS1)
Databases: ClinVar
ClinGen SVI recommended retiring PP5; retained for maximum sensitivity.
Benign Evidence (12 criteria)
Allele frequency above 5%
Conditions
gnomAD global AF > 0.05 (5%)
Databases: gnomAD v4.1
BA1 overrides ALL other evidence including ClinVar. VCEP gene-specific BA1 thresholds may be lower.
Allele frequency greater than expected for disorder
Conditions
8-level inheritance-aware cascade: VCEP > ClinGen HI=3 > Orphanet AD > Orphanet XLD > HPO AD > Orphanet AR > AR_LOF_GENES > Constraint-implied AD > Default AR
AD threshold: AF >= 0.1%
AR threshold: AF >= 5%
Upper bound: AF <= BA1 threshold
Databases: gnomAD v4.1, ClinGen (HI score), Orphanet (inheritance), HPO (HP:0000006), gnomAD Constraint (pLI, LOEUF), AR_LOF_GENES
VCEP gene-specific BS1 thresholds override generic logic when enabled
Observed in healthy adults for fully penetrant early-onset disorder
Conditions
Inheritance-aware threshold (v3.26.8, HELIX-CR-2026-074): AR-only genes (Orphanet has_ar=TRUE, has_ad=FALSE OR ClinGen HI=30) with documented early-onset-only diseases use threshold of 10 homozygotes. Generic and AD genes use threshold of 15 homozygotes
Onset guard from refdb.orphanet_disease_onset (HELIX-REF-005): AR threshold activated only when gene exists in onset table AND no associated disease has Adult/Elderly/All ages onset. Aligns with Richards 2015 Table 3: BS2 "with full penetrance expected at an early age"
AR path criteria string marker: BS2[AR] when AR threshold path was used
Exclusions
AD BS2 from gnomAD homozygote count not implemented (Harrison 2019 PMID:31159682; APC VCEP Curia 2023 PMID:37805481 excludes gnomAD for AD BS2)
X-linked hemizygous BS2 deferred (ac_hemi column not in current annotation pipeline)
Databases: gnomAD v4.1 (global_hom), Orphanet (orphanet_gene_inheritance, orphanet_disease_onset), ClinGen (HI=30 AR proxy)
AR threshold of 10 calibrated for early-onset full-penetrance AR diseases. Adult-onset AR diseases (e.g., Wilson disease ATP7B) fall back to generic 15 via onset guard
Functional studies show no deleterious effect
Requires curation of published functional assay data.
Lack of segregation in affected family members
Requires family segregation data.
Missense in a gene where primarily truncating variants cause disease
Conditions
Missense variant
MODERATE impact
pLI < 0.1 (LoF-tolerant gene)
mis_z < 2.0 or absent (v3.6.8 missense constraint guard)
No session-local ClinVar P/LP missense in same gene (v3.8.2 guard)
Reference-based ClinVar missense guard (v3.20.9, HELIX-CR-2026-044): blocked when gene is in refdb.clinvar_missense_genes (>= 2 ClinVar P/LP missense at 2+ stars). Reference-level check covers genes with established missense disease mechanism not present in current patient VCF
Exclusions
GoF / DN gene exclusion (v3.20.6, HELIX-CR-2026-039v2): BP1 not applied for missense in gain-of-function or dominant-negative genes. BP1 presupposes LoF mechanism per Richards 2015 Table 3; GoF/DN genes have non-LoF mechanism, so the BP1 premise is inapplicable. Same guard sources as PVS1 GoF guard: refdb.gof_genes_exclusive + GOF_AD_GENES fallback
Curated AR LoF genes (~150 genes with biallelic LoF disease mechanism)
Databases: VEP, gnomAD Constraint (pLI, mis_z), ClinVar (session-local), refdb.clinvar_missense_genes (reference-level), refdb.gof_genes_exclusive, GOF_AD_GENES, AR_LOF_GENES
mis_z measures heterozygous missense constraint. AR enzyme genes (e.g., MCCC2) may have low mis_z despite pathogenic homozygous missense -- ClinVar missense guards (session-local v3.8.2 + reference-based v3.20.9) address this gap
Observed in trans with pathogenic for fully penetrant dominant
Conditions
Compound heterozygote candidate
ClinGen haploinsufficiency score = 30 (dosage sensitivity unlikely)
Databases: ClinGen
In-frame indel in repetitive region without known function
Conditions
In-frame indel
Repetitive/low-complexity region OR not in any Pfam domain
Databases: VEP (consequence, domains)
Computational evidence suggests no impact
Conditions
Path A (Missense -- BayesDel): consequence = missense_variant required (v3.12.1, HELIX-CR-2026-009). BayesDel_noAF with ClinGen SVI calibrated thresholds (Pejaver 2022) -- BP4_Moderate (<= -0.361, -2 pts), BP4_Supporting (-0.360 to -0.181, -1 pt). SpliceAI max_score must be < 0.1
Path B (Non-canonical splice -- BP4_splice): v3.25.2 (HELIX-CR-2026-060). Supporting benign for non-canonical splice region variants (splice_donor_region, splice_donor_5th_base, splice_acceptor_5th_base, splice_region, splice_polypyrimidine_tract) when SpliceAI max_score <= 0.1. Excludes synonymous (BP7 handles) and missense (BP4 BayesDel handles). Threshold consistent with BP7. References: Jaganathan 2019 PMID:30661751, Walker 2023 PMID:37352859, de la Hoya 2022 PMID:35202600
Criteria string markers: BP4 (BayesDel) and BP4_splice (non-canonical splice) reported separately
Exclusions
BayesDel indeterminate range (-0.180 to 0.129): no BP4 missense path applied
BP4_splice not applied when PVS1 is triggered (canonical splice +/-1,2 with constraint pass) -- prevents double-counting splice/LoF evidence
Databases: dbNSFP 4.9c (bayesdel_noaf_score), SpliceAI (max_score), VEP (consequence)
BP4_splice covers only non-canonical splice region consequences. Deep intronic variants without VEP splice tags require RNA validation per ClinGen SVI (Walker 2023)
Variant found in case with alternate molecular basis
Requires case-level information about alternative diagnoses.
Reputable source reports variant as benign
Conditions
ClinVar: Benign or Likely Benign
Review stars >= 1
Databases: ClinVar
ClinGen SVI recommended retiring BP6; retained for maximum sensitivity.
Synonymous variant with no predicted splice impact
Conditions
Synonymous variant
Not in splice region
SpliceAI max_score <= 0.1
Databases: VEP, SpliceAI
Conservation filter intentionally omitted per Walker et al. 2023 Table S13.