ACMG Assignment
Starting with version 13.1.0, Exomiser performs a partial categorisation of the variants contributing to the gene score for a mode of inheritance using the ACMG/AMP Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. The criteria are assigned according to the UK ACGS 2020 guidelines and scored according to the ClinGen Variant Classification Guidance updated 2020 guidelines.
Additionally, splice variants are scored according to Using the ACMG/AMP framework to capture evidence related to predicted and observed impact on splicing: Recommendations from the ClinGen SVI Splicing Subgroup. In this case Exomiser will assign the PVS1, PS1, PP3, BP4, BP7 categories.
It is important to be aware that these scores are not a substitute for manual assignment by a qualified clinical geneticist or clinician - The scores displayed utilise the data found in the Exomiser database and are a subset of the possible criteria by which to assess a variant. Nonetheless, in our benchmarking on the returned cases from the 100K Genomes Project, restricting to variants with these automated P/LP classifications increases precision (positive predictive value) markedly without excluding many real diagnoses. For example, on a cohort of 742 solved cases where the top 5 Exomiser candidates were considered, including the P/LP criteria increased precision 3.8-fold from 15% to 57% with only a small drop in the recall of the diagnoses from 94% to 83%. An even larger 5.7-fold increase of precision from 3% to 17% was observed when considering a larger cohort of 31k cases where only 17% had received a positive diagnosis (again with a modest drop in recall from 91% to 75%).
Exomiser is capable of assigning the following ACMG categories:
Variant Classification Evidence
These categories are taken from Figure 1. of Richards et al. 2015, those in bold are assigned by Exomiser.
Population Data
For the population data criteria, all frequencies are considered using the populations set by the user in the frequencySources:, apart from any bottle-necked populations not recommended for frequency filtering from gnomAD according to their filtering allele frequency document. This excludes the Ashkenazi Jewish (ASJ), European Finnish (FIN), Other (OTH), Amish (AMI) and Middle Eastern (MID) populations. In addition the LOCAL frequency will also not be used.
BA1
MAF is too high for disorder
Given Exomiser will filter out alleles with an allele frequency of >= 2.0%, this is unlikely to be seen. However, alleles with a maximum frequency >= 5.0% in the frequency sources specified will be assigned the BA1 criterion. Variants listed as being excluded from this category by the ClinGen SVI working group BA1 exclusion list will not be marked as BA1, assuming they survived variant filtering.
BS1
MAF is too high for disorder
Assigned to autosomal variants with a maximum non-founder population allele frequency >= 1.5% or mitochondrial variants with MAF >= 5%
BS2
Observation in controls inconsistent with disease penetrance
Variants with a MAF < 5% (not BA1) which have a gnomAD non-founder population hom count > 5 when considered under an autosomal dominant model (gene score) or > 2 when considered under autosomal recessive, X-recessive or X-dominant are assigned a BS2.
PM2
Absent in population databases
In accordance with the updated PM2 guidance, variants absent from all of the user-defined frequencySources: will be assigned the PM2_Supporting criterion. Additionally, for variants considered under a recessive mode of inheritance they can have a frequency of < 0.01% (0.0001) in all non-bottlenecked populations to be assigned PM2_Supporting.
PS4
Prevalence in affecteds statistically increased over controls
Not assigned
Computational and Predictive Data
PVS1
Predicted null variant in a gene where LOF is a known mechanism of disease
Variants must have a predicted loss of function effect, be in a gene with known disease associations and have a gene constraint LOF O/E < 0.7635 (gnomAD 4.0) to suggest that a gene is LoF intolerant. Variants not predicted to lead to NMD (those located in the last exon) will have the modifier downgraded to Strong.
PS1
Same amino acid change as an established pathogenic variant
Variants with the same amino acid change as previously reported P/LP missense or in-frame indel ClinVar variants will be assigned PS1 with a strength of Strong for variants >= 2 stars, Moderate for variants with 1 star or Supporting for those without a ClinVar start rating.
Splice variants will be assigned PS1 with Strong, Moderate or Supporting modifiers, according to table 2 of Using the ACMG/AMP framework to capture evidence related to predicted and observed impact on splicing: Recommendations from the ClinGen SVI Splicing Subgroup.
PM4
Protein length changing variant
Stop-loss and in-frame insertions or deletions, not previously assigned a PVS1 criterion are assigned PM4.
PM5
Novel missense change at an amino acid residue where a different pathogenic missense change has been seen before
Variants having a novel missense change to an amino acid where a previously reported ClinVar P/LP variant has been seen will be assigned PM5 with a strength of Moderate for those with >=2 stars or Supporting otherwise.
PP3 / BP4
Multiple lines of computational evidence support a deleterious effect on the gene/gene product (PP3) Multiple lines of computational evidence suggest no impact on gene product (BP4)
If REVEL is chosen as a pathogenicity predictor for missense variants, PP3 and BP4 are assigned using the modifiers according to table 2 of Evidence-based calibration of computational tools for missense variant pathogenicity classification and ClinGen recommendations for clinical use of PP3/BP4 criteria. Note that this suggests the use of modifiers up to Strong in the case of pathogenic or Very Strong in the case of benign predictions. Otherwise, an ensemble-based approach will be used for other pathogenicity predictors as per the original 215 guidelines. It should be noted we found better performance using the REVEL-based approach when testing against the 100K genomes data.
For splice variants outside of the +/-1,2 canonical splice donor/acceptor site are assigned PP3 with a SpliceAI score >= 0.2 or BP4 with a SpliceAI score < 0.1
BP1
Missense in gene where only truncating cause disease
Missense or inframe indels found in a gene in which >=75% of ClinVar P/LP variants are loss of function variants are assigned BP1.
BP7
Silent variant with non predicted splice impact
For synonymous variants, BP7 is assigned if the variant is not reported as P/LP in ClinVar, with a 1+ star rating and has a SpliceAI score < 0.1
For splice region variants Exomiser follows the recommendations made in Using the ACMG/AMP framework to capture evidence related to predicted and observed impact on splicing: Recommendations from the ClinGen SVI Splicing Subgroup
In this case, non-donor/acceptor splice region variants with a SpliceAI score < 0.1 which have been assigned BP4 at or beyond the +7 and −21 positions (conservative designation for donor/acceptor splice region) are also assigned a BP7.
Functional Data
PS3
Well-established functional studies show a deleterious effect
Not assigned
PM1
Mutational hot spot or well-studied functional domain without benign variation
Missense and inframe indels are assigned PM1 if the surrounding region of 25 nucleotides either side of the variant contain at least 4 reported P/LP variants in ClinVar and no B/LB variants. If the number of P/LP variants is greater than the number of VUS in the region the strength will be assigned Moderate but regions containing P/LP <= VUS (and no B/BL) will have the strength downgraded to Supporting.
PP2
Missense in gene with low rate of benign missense variants and path. missenses common
Missense or inframe indels in genes where >= 75% of ClinVar P/LP variants are missense and ClinVar benign missense variants make up <=5% of all reported missense ClinVar variants (P/LP/VUS/B/LB) in that gene are assigned a PP2
BS3
Well-established functional studies show no deleterious effect
Not assigned
Segregation Data
PP1
Co-segregation with disease in multiple affected family members
Not assigned
BS4
Non-segregation with disease
Not assigned. Previously this was assigned but the pedigree assignment of ‘affected’ status was not always reliable so this is no longer applied automatically.
De novo Data
PS2
*De novo* (paternity & maternity confirmed)
Exomiser assigns the PS2 criterion for variants compatible with a dominant mode of inheritance, with a pedigree containing at least two ancestors of the proband, none of whom are affected and none of whom share the same allele as the proband.
PM6
*De novo* (without paternity & maternity confirmed)
Not assigned
Allelic Data
PM3 / BP2
Observed in *trans* with a dominant variant Observed in *cis* with a pathogenic variant
For recessive disorders, detected in trans with a pathogenic variant PM3
If Exomiser is provided with a phased VCF and a variant is found to be in-trans with a ClinVar Pathogenic variant and associated with a recessive disorder, the PM3 criterion will be applied. However, in cases where variant is being considered for a recessive disorder and is in-cis or a dominant disorder and in-trans with another pathogenic variant the BP2 criterion is applied.
Phenotype
PP4
Patient’s phenotype or FH highly specific for gene
Given Exomiser’s focus on phenotype-driven variant prioritisation, variants in a gene associated with a disorder with a phenotype match score > 0.6 to the patient’s phenotype are assigned the PP4 criterion at the Moderate, rather than Supporting level.
BP5
Found in case with an alternate cause
Not assigned
Clinical
PP5 / BP6
Reputable source w/out shared data = benign (BP6) Reputable source = pathogenic (PP5)
If a variant is previously reported as P/LP in ClinVar with a 1-start rating, it will be assigned PP5, those with >= 2 stars (multiple submitters, criteria provided, no conflicts / reviewed by expert panel / practice guideline) will be assigned a Strong level. Conversely, if the variant is previously reported as B/LB it will be assigned BP6 with the same modification criteria. Typically these P/LP variants will be in the Exomiser ClinVar ‘whitelist’, and will have a very high variant score irrespective of the predicted variant effect and always survive any filtering criteria.
Transcript Selection
Transcripts will be selected using the most deleterious predicted variant effect from Jannovar according to the transcript-source property set in the application.properties. We recommend using the Ensembl transcript datasource as the Exomiser build uses the GENCODE basic transcripts for hg19/GRCh37 and MANE transcripts for hg38/GRCh38.
ACMG assignments will be reported for a variant on a transcript consistent with a particular mode of inheritance in conjunction with a disorder, the assigned criteria with any modifiers and the final classification e.g.
1-12335-A-T, NC_000001.10:g.12335A>T, GENE1(ENST12345678):c.2346A>T:p.1234A>-, PATHOGENIC, [PVS1, PS1, PP4_Strong], Disease (OMIM:12345), AUTOSOMAL_DOMINANT
"acmgAssignments": [
{
"variantEvaluation": {
"genomeAssembly": "HG19",
"contigName": "10",
"start": 123256215,
"end": 123256215,
"ref": "T",
"alt": "G",
"type": "SNV",
"length": 1,
"phredScore": 100,
"variantEffect": "MISSENSE_VARIANT",
"whiteListed": true,
"filterStatus": "PASSED",
"contributesToGeneScore": true,
"variantScore": 1,
"frequencyScore": 1,
"pathogenicityScore": 1,
"predictedPathogenic": true,
"passedFilterTypes": [
"FAILED_VARIANT_FILTER",
"PATHOGENICITY_FILTER",
"FREQUENCY_FILTER",
"VARIANT_EFFECT_FILTER",
"INHERITANCE_FILTER"
],
"frequencyData": {
"rsId": "rs121918506",
"frequencyScore": 1
},
"pathogenicityData": {
"clinVarData": {
"alleleId": "28333",
"primaryInterpretation": "LIKELY_PATHOGENIC",
"reviewStatus": "criteria provided, single submitter"
},
"pathogenicitycore": 0.965,
"pathogenicityScores": [
{
"source": "REVEL",
"score": 0.965
},
{
"source": "MVP",
"score": 0.9517972
}
],
"mostPathogenicScore": {
"source": "REVEL",
"score": 0.965
}
},
"compatibleInheritanceModes": [
"AUTOSOMAL_DOMINANT"
],
"contributingInheritanceModes": [
"AUTOSOMAL_DOMINANT"
],
"transcriptAnnotations": [
{
"variantEffect": "MISSENSE_VARIANT",
"geneSymbol": "FGFR2",
"accession": "ENST00000346997.2",
"hgvsGenomic": "g.12278533A>C",
"hgvsCdna": "c.1688A>C",
"hgvsProtein": "p.(Glu563Ala)",
"rankType": "EXON",
"rank": 12,
"rankTotal": 17
},
{
"variantEffect": "MISSENSE_VARIANT",
"geneSymbol": "FGFR2",
"accession": "ENST00000351936.6",
"hgvsGenomic": "g.12278533A>C",
"hgvsCdna": "c.1688A>C",
"hgvsProtein": "p.(Glu563Ala)",
"rankType": "EXON",
"rank": 13,
"rankTotal": 18
}
]
},
"geneIdentifier": {
"geneId": "ENSG00000066468",
"geneSymbol": "FGFR2",
"hgncId": "HGNC:3689",
"hgncSymbol": "FGFR2",
"entrezId": "2263",
"ensemblId": "ENSG00000066468",
"ucscId": "uc057wle.1"
},
"modeOfInheritance": "AUTOSOMAL_DOMINANT",
"disease": {
"diseaseId": "OMIM:123150",
"diseaseName": "Jackson-Weiss syndrome",
"associatedGeneId": 2263,
"diseaseType": "DISEASE",
"inheritanceMode": "AUTOSOMAL_DOMINANT",
"phenotypeIds": [
"HP:0000006",
"HP:0000272",
"HP:0001363",
"HP:0001783",
"HP:0004691",
"HP:0008080",
"HP:0008122",
"HP:0010055",
"HP:0010743",
"HP:0011800"
],
"id": "OMIM:123150",
"associatedGeneSymbol": "FGFR2"
},
"acmgEvidence": {
"evidence": {
"PM2": "MODERATE",
"PP3": "STRONG",
"PP4": "SUPPORTING",
"PP5": "SUPPORTING"
}
},
"acmgClassification": "LIKELY_PATHOGENIC"
}
]
}