Criteria combination rules

Last updated: August 29, 2026.

ACMG/AMP criteria are not independent evidence by definition. Before summing two criteria, you must verify that they represent biologically independent lines of evidence — the same molecular, population, computational, functional, phenotypic, or family observation must not be counted twice.

VariClass warns (without blocking) when two criteria marked together match one of the combinations below, with the justification and specific reference. The final decision always rests with the evaluator.

Do not combine

PVS1 + PM1

PVS1 and PM1 must not be combined for the same variant.

PVS1 represents loss of function in a known LOF mechanism; PM1 represents location in a hotspot/critical functional domain. ACGS 2024 states they must not be used together.

Source: ACGS 2024, p. 10

PVS1 + PM4

PM4 must not be applied if PVS1 is already in use.

PVS1 and PM4 can represent alternative interpretations of the same structural consequence (e.g., stop-loss). Do not sum both criteria for the same consequence.

Source: ACGS 2024, pp. 9–10, 13

PVS1 + PP2

PP2 must not be applied if PVS1 is already in use.

ACGS 2024 lists PP2 among the criteria that must not be used with PVS1; PP2's missense-restriction reasoning generally adds no independent evidence to a variant already classified as LOF by PVS1.

Source: ACGS 2024, p. 10

PVS1 + PP3

PVS1 and PP3 must not be used together.

For splicing variants, computational prediction is often already integrated into the assessment that precedes the determination of the LOF effect; it should not generate additional independent weight when PVS1 already applies.

Source: ACGS 2024, p. 10

PVS1 + PS3

PS3 must not be added when PVS1 is applied at Very Strong strength.

For variants without NMD, functional data can contribute to raising PVS1's weight up to Very Strong; in that scenario PS3 must not be added separately as a second piece of evidence.

Source: ACGS 2024, p. 10

PS1 + PM4

PS1 must not be used together with PM4.

ACGS 2024 establishes this restriction explicitly in the PS1 refinement.

Source: ACGS 2024, p. 10

PS2 + PM6

PS2 and PM6 must not be summed for the same de novo event.

They are alternative representations of the strength of the same type of evidence (PS2: confirmed parentage; PM6: assumed de novo without formal confirmation) — choose one, not both.

Source: ACGS 2024, pp. 6–7, 10–11; ClinGen SVI de novo

Do not reuse the same evidence

PM4 + PP3

Do not reuse the same computational prediction to support both PM4 and PP3.

If in silico information has already been incorporated into the PM4 decision, it must not be presented again as an independent PP3.

Source: ACGS 2024, p. 13

BP3 + BP4

Do not reuse the same in silico evidence for BP3 and BP4.

If computational tools were used as part of the rationale for BP3, they must not be counted again as BP4.

Source: ACGS 2024, pp. 16–17

PS2 + PP4

The same patient's phenotypic specificity used for PS2 must not be added again as PP4.

The strength of the de novo criterion is already modulated by the consistency/specificity of the phenotype; counting the same phenotype a second time in PP4 creates dependency.

Source: ACGS 2024, pp. 7, 10–11

PM6 + PP4

The same patient's phenotypic specificity used for PM6 must not be added again as PP4.

The strength of the de novo criterion is already modulated by the consistency/specificity of the phenotype; counting the same phenotype a second time in PP4 creates dependency.

Source: ACGS 2024, pp. 7, 10–11

PP4 + PS4

Do not reuse the same clinical cases for PS4 and PP4.

Avoid double-counting the same cases between case enrichment (PS4) and phenotypic specificity (PP4).

Source: ACGS 2024, pp. 15–16

PP1 + PS4

Do not reuse the same relatives/cases between PP1 and PS4.

Avoid the same family increasing the weight twice — through segregation (PP1) and through cases (PS4).

Source: ACGS 2020/2024

Require independent evidence

PM1 + PM5

PM1 and PM5 may only coexist if supported by independent evidence.

They are not incompatible by definition, but must not share the same observation — e.g., the same pathogenic variant cannot simultaneously be the sole reason for PM1 (hotspot) and the basis for PM5 (same residue).

Source: ACGS 2024, pp. 12–13

PM1 + PP2

Avoid double-counting constraint between PM1 and PP2.

If the same regional constraint metric (missense variation intolerance) supports both criteria, the evidence is not independent.

Source: ACGS 2024, pp. 12, 14

Documented preference

PM3 + PS4

For autosomal recessive disorders, prefer PM3 over PS4 for the same rare cases.

PM3 specifically represents observation in trans/homozygosity in the recessive model; using the same cases in PS4 can duplicate evidence. Exception: common recessive variants where PM2 does not apply may justify PS4.

Source: ACGS 2024, p. 13

Manual review recommended

PS1 + PM5

Verify independence: both rely on previously established amino acid substitutions.

There is potential dependency, but no general explicit prohibition in the sources used — do not block automatically without a specific VCEP rule.

Source: ACGS 2024

PS1 + PM1

Check whether PM1 is supported solely by the same variant used for PS1.

If PM1 depends exclusively on PS1's reference variant, there is double counting; if it is supported by a robust and independent hotspot/domain, there is no general prohibition.

Source: ACGS 2024

PS3 + PM1

Check whether PS3's functional assay is the only reason for declaring the critical domain/residue in PM1.

A variant-specific assay may be independent of the regional/domain evidence; avoid combining only in this overlap scenario.

Source: ACGS 2024

BS2 + BP2

Check whether the same healthy individual/genotypic configuration supports both criteria.

There may be dependency; the sources used do not provide a general prohibition equivalent to PVS1's explicit rules.

Source: ACGS 2024

BP2 + BP5

Check whether the same alternate pathogenic variant supports both criteria.

Double counting can occur if the same alternate variant is used in BP2 (co-occurrence) and in BP5 (alternate molecular cause).

Source: ACGS 2024, pp. 16–17

Explicitly allowed combinations

PS3 + PP3

Allowed for missense variants when the functional assay and the computational prediction are independent.

Source: ACGS 2024, p. 11

BS1 + BS2

Explicitly allowed, provided each criterion is individually valid.

Source: ACGS 2024, p. 16

BP4 + BP7

Allowed for synonymous variants with no predicted splicing effect, when no RNA evidence is being used for BP7.

Source: ACGS 2024, pp. 16–17

What is not checked automatically

Some guideline rules depend on context that VariClass does not collect — for example, whether a PP3, BP4, PS3, or BP7 refers specifically to splicing and whether RNA evidence is involved (PVS1[RNA], BP7[RNA]), or the structured provenance of each piece of evidence (same case, same family, same assay). These rules do not trigger an automatic alert; the exclusion of PVS1 + PP3 and the default allowance of BP4 + BP7 already cover the most common scenarios in this category, but reviewing the context remains the evaluator's responsibility.

References

Durkie M, Cassidy EJ, Berry I, Owens M, Turnbull C, Scott RH, et al. ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2024. Association for Clinical Genomic Science; 2024. Version 1.2.
Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of ACMG and AMP. Genet Med. 2015;17(5):405-424.
Tavtigian SV, Greenblatt MS, Harrison SM, Nussbaum RL, Prabhu SA, Boucher KM, Biesecker LG. Modeling the ACMG/AMP variant classification guidelines as a Bayesian classification framework. Genet Med. 2018;20(9):1054-1060.
Abou Tayoun AN, Pesaran T, DiStefano MT, Oza A, Rehm HL, Biesecker LG, Harrison SM. Recommendations for interpreting the loss of function PVS1 ACMG/AMP variant criterion. Hum Mutat. 2018;39(11):1517-1524.
Brnich SE, Abou Tayoun AN, Couch FJ, Cutting GR, Greenblatt MS, Heinen CD, et al. Recommendations for application of the functional evidence PS3/BS3 criterion using the ACMG/AMP sequence variant interpretation framework. Genome Med. 2020;12:3.