Table 4

Permissible and non-permissible combinations of concordant evidence elements (CanVIG-UK consensus)

ThemeEvidence elementsCanVIG consensusNotes, references
In silico + functional dataPS3 (functional) and PP3 (in silico)XCo-usage permitted for assays of protein function
Co-usage not permitted for assays of splicing; in silico evidence incorporated into PS3 for assays of splicing (as per ACGS guidance)4 5 19 21 22
BS3 (functional) and BP4 (in silico)X
PM1 (hot spot) and PP3 (in silico)Co-usage permitted; regional enrichment and in silico prediction largely different evidence types (and evidence from three tools generally incorporated)15
In silico + variant mechanismBP4 (in silico) and BP7 (silent variant)Co-usage permitted for synonymous and intronic variants (splicing effect must be excluded)
PVS1 (null variant) and PP3 (in silico)XCo-usage not permitted; in silico predictors often incorporate variant mechanism. The strongest legitimate evidence item should be selected for inclusion.23
PM4 (protein length changes) and PP3 (in silico)X
BP3 (in-frame del/ins) & BP4 (in silico)X
Use of population dataPS4 (case control) and PM2 (absence in controls)Co-usage permitted, provided that a different source of population data is used for each (this can comprise two predefined partitions of gnomAD)5
Phenotypic specificityPS4 (case control) and PP4 (phenotype specificity)Co-usage permitted, provided that there is a schema predefining distinct data types used for each, thus preventing ‘double counting’ of phenotypic features (eg PP4 is applied for molecular/tumour assay data, indicating specificity at gene level); PS4 is applied at patient level, counting the strength of features in cases and the number of cases/families5
Scaled evidence itemsPS2 (de novo confirmed) and PM6 (de novo not confirmed)XCo-usage not permitted; either PS2 or PM6 is used, depending on aggregate observations of de novo occurrence (ClinGen Sequence Variant Interpretation Working Group)24
Overlapping/ related evidencePS1 (same amino acid) and PM5 (same residue)XCo-usage not permitted; the strongest legitimate evidence item should be selected for inclusion
PM1 (hot spot) and PP2 (low missense rate)Co-usage permitted; there may be both a low rate of benign missense variation at a whole gene/gene region level and a specific mutational hot spot/functional domain
PVS1 (null variant) and PM4 (protein length changes)XCo-usage not permitted; this is double counting of two end points of deleterious effect23
PVS1 (null variant) and PS3 (functional)XCo-usage not permitted for canonical splice variants or non-canonical splice variants (experimental evidence cannot be scored using PS3 if PVS1 is used)23
Co-usage is permitted for assays of protein function where the evidence is not double-counted (eg, truncating variant in last exon of gene scored as PVS1_mod can be combined with PS3 for experimental evidence, demonstrating a significant effect on protein function)
PS1 (same amino acid) and PM1 (hot spot)✔*Co-usage permitted; definition of a hot spot is predicated on multiple well-documented pre-existing pathogenic variants. This may include those at the same amino acid
PM5 (same residue) and PM1 (hot spot)✔*
Alternative explanation for phenotypeBP2 (observed with pathogenic variant) and BP5 (alternative cause)XCo-usage not permitted; These evidence elements both pertain to presence of an alternative genetic cause (BP2 in the same gene and BP5 in a different gene)
  • *PM1 should therefore be used at supporting level if used in combination with PS1 or PM5.15

  • ACGS, Association for Clinical Genomic Science; CanVIG-UK, Cancer Variant Interpretation Group UK.