ACMG/AMP 2015 — sequence variant classification rules
The ACMG/AMP 2015 germline sequence-variant classification framework — the 16 pathogenic evidence codes (PVS1, PS1–PS4, PM1–PM6, PP1–PP5), the 12 benign codes (BA1, BS1–BS4, BP1–BP7) and the Table 5 combining rules that map them onto pathogenic, likely pathogenic, uncertain significance, likely benign and benign — the rule set a clinical molecular geneticist applies to every variant before it is reported.
Rendering…
Make it your own.
digraph acmg {
rankdir=LR;
graph [fontname="Helvetica", fontsize=11, nodesep=0.3, ranksep=0.75, bgcolor="transparent",
label="ACMG/AMP 2015 sequence-variant classification (Richards et al., Genet Med 17:405) — the evidence codes and the rules that combine them.\lApply to germline variants in Mendelian disease. Every reported variant carries one of the five classes; a class is defensible only if the codes that produced it are listed.\l",
labelloc="t", labeljust="l"];
node [fontname="Helvetica", fontsize=9, shape=box, style="filled", fillcolor="#f8fafc", color="#64748b", margin="0.14,0.09"];
edge [fontname="Helvetica", fontsize=8, color="#94a3b8"];
subgraph cluster_path {
label="Evidence for pathogenicity"; style="rounded,filled"; fillcolor="#fef2f2"; color="#fca5a5"; fontsize=11; fontcolor="#b91c1c";
pvs [label="VERY STRONG\lPVS1 null variant (nonsense, frameshift, canonical ±1/2 splice,\l initiation codon, single- or multi-exon deletion) in a gene\l where loss of function is a known disease mechanism\l", fillcolor="#fee2e2"];
ps [label="STRONG\lPS1 same amino-acid change as an established pathogenic variant\lPS2 de novo, maternity and paternity confirmed\lPS3 well-established functional assay shows a damaging effect\lPS4 prevalence in affected significantly increased over controls\l", fillcolor="#fee2e2"];
pm [label="MODERATE\lPM1 mutational hot spot / critical functional domain\lPM2 absent from gnomAD (or very low frequency if recessive)\lPM3 in trans with a pathogenic variant (recessive disorder)\lPM4 protein length change: in-frame indel or stop-loss\lPM5 novel missense at a residue with another pathogenic missense\lPM6 assumed de novo, parentage not confirmed\l", fillcolor="#fee2e2"];
pp [label="SUPPORTING\lPP1 co-segregation with disease in multiple affected relatives\lPP2 missense in a gene with low benign missense variation\lPP3 multiple computational lines support a deleterious effect\lPP4 phenotype highly specific for a single-gene disease\lPP5 reputable source reports pathogenic (retired, ClinGen SVI 2018)\l", fillcolor="#fee2e2"];
}
subgraph cluster_benign {
label="Evidence for a benign role"; style="rounded,filled"; fillcolor="#eff6ff"; color="#93c5fd"; fontsize=11; fontcolor="#1d4ed8";
ba [label="STAND-ALONE\lBA1 allele frequency > 5% in gnomAD / 1000G / ESP\l", fillcolor="#dbeafe"];
bs [label="STRONG\lBS1 allele frequency greater than expected for the disorder\lBS2 seen in a healthy adult where full penetrance is expected\lBS3 well-established functional assay shows no damaging effect\lBS4 lack of segregation in affected family members\l", fillcolor="#dbeafe"];
bp [label="SUPPORTING\lBP1 missense in a gene where only truncating variants cause disease\lBP2 in trans with a pathogenic variant in a fully penetrant\l dominant gene, or in cis with a pathogenic variant\lBP3 in-frame indel in a repeat region of unknown function\lBP4 multiple computational lines suggest no impact\lBP5 an alternate molecular basis for the disease was found\lBP6 reputable source reports benign (retired, ClinGen SVI 2018)\lBP7 synonymous, no predicted splice effect, nucleotide not conserved\l", fillcolor="#dbeafe"];
}
subgraph cluster_rules {
label="Combining rules — ACMG/AMP 2015 Table 5"; style="rounded,filled"; fillcolor="#f8fafc"; color="#cbd5e1"; fontsize=11; fontcolor="#334155";
rp [label="PATHOGENIC if any one holds\l(i) 1 PVS1 AND ≥1 strong OR ≥2 moderate\l OR 1 moderate + 1 supporting OR ≥2 supporting\l(ii) ≥2 strong\l(iii) 1 strong AND (≥3 moderate) OR (2 moderate + ≥2 supporting)\l OR (1 moderate + ≥4 supporting)\l", fillcolor="#fee2e2", color="#dc2626"];
rlp [label="LIKELY PATHOGENIC if any one holds\l(i) 1 PVS1 + 1 moderate\l(ii) 1 strong + 1–2 moderate\l(iii) 1 strong + ≥2 supporting\l(iv) ≥3 moderate\l(v) 2 moderate + ≥2 supporting\l(vi) 1 moderate + ≥4 supporting\l", fillcolor="#ffedd5", color="#ea580c"];
rb [label="BENIGN if either holds\l(i) 1 BA1\l(ii) ≥2 strong (BS1–BS4)\l", fillcolor="#dbeafe", color="#2563eb"];
rlb [label="LIKELY BENIGN if either holds\l(i) 1 strong + 1 supporting\l(ii) ≥2 supporting (BP1–BP7)\l", fillcolor="#e0f2fe", color="#0284c7"];
rvus [label="UNCERTAIN SIGNIFICANCE\l(i) no combining rule above is met, or\l(ii) benign and pathogenic criteria are both met and contradict\l", fillcolor="#f1f5f9", color="#64748b"];
}
P [label="PATHOGENIC\nreport; offer cascade testing", shape=box, style="rounded,filled", fillcolor="#dc2626", fontcolor="white", fontsize=11];
LP [label="LIKELY PATHOGENIC\n≥ 90% certainty; report", shape=box, style="rounded,filled", fillcolor="#ea580c", fontcolor="white", fontsize=11];
VUS [label="VUS\ndo not use for clinical decisions;\nre-evaluate as evidence accrues", shape=box, style="rounded,filled", fillcolor="#94a3b8", fontcolor="white", fontsize=11];
LB [label="LIKELY BENIGN\n≥ 90% certainty", shape=box, style="rounded,filled", fillcolor="#0284c7", fontcolor="white", fontsize=11];
B [label="BENIGN", shape=box, style="rounded,filled", fillcolor="#2563eb", fontcolor="white", fontsize=11];
pvs -> rp; ps -> rp; pm -> rp; pp -> rp;
pvs -> rlp; ps -> rlp; pm -> rlp; pp -> rlp;
ba -> rb; bs -> rb;
bs -> rlb; bp -> rlb;
rp -> P;
rlp -> LP;
rb -> B;
rlb -> LB;
rvus -> VUS;
rp -> rvus [label=" criteria not met", style=dashed];
rb -> rvus [label=" conflicting evidence", style=dashed];
}