How Does LYFE Sciences Apply ACMG Criteria?

BP4, BP7, and PP3

We use a combination of computational tools to evaluate genetic variants and apply the ACMG criteria BP4, BP7, and PP3.

For the REVEL score, a value less than 0.25 is considered benign, greater than 0.75 is pathogenic, and scores in between are classified as uncertain.

For the SIFT score, scores between 0.0 and 0.05 are pathogenic, and between 0.05 and 1.0 are benign.

For the PolyPhen-2 score, scores close to 1.0 indicate probable damage.

For the SIFT4G score, scores less than 0.05 are considered pathogenic.

For the LRT score, 'D' (Deleterious) is considered pathogenic, 'N' (Neutral) is considered benign, and 'U' (Unknown) is classified as uncertain.

For the MutationTaster score, scores greater than 0.5 are considered pathogenic.

For the MutationAssessor score, high scores (e.g., 2.805) indicate pathogenicity.

For the FATHMM score, scores less than or equal to -1.5 indicate pathogenicity.

For the PROVEAN score, scores less than or equal to -2.5 indicate pathogenicity.

For the MetaSVM score, scores greater than 0 indicate pathogenicity.

For the PrimateAI score, scores greater than 0.803 indicate pathogenicity.

For the MetaLR score, scores greater than 0.5 indicate pathogenicity.

For the MetaRNN score, scores greater than 0.5 indicate pathogenicity.

For the BayesDel AddAF score, scores greater than 0.0692655 indicate pathogenicity.

For the BayesDel NoAF score, scores greater than -0.0570105 indicate pathogenicity.

For the FATHMM-MKL score, scores greater than 0.5 indicate pathogenicity.

For the FATHMM-XF score, scores greater than 0.5 indicate pathogenicity.

We use dbNSFP¹, a database of pathogenicity predictions for missense variants. See dbNSFP README for a detailed description of predictors and how they are evaluated

Final Criteria

The final criteria are determined based on the majority of predictor scores:

• BP4 or BP7: If benign predictors outnumber pathogenic and uncertain significance (VUS) predictors.
• PP3: If pathogenic predictors outnumber benign and VUS predictors.
• VUS: If neither benign nor pathogenic predictors predominate, the variant is classified as VUS.

PM1

We apply the ACMG criterion PM1 to genetic variants that are located in mutational hot spots or well-studied functional domains with evidence of a damaging mechanism. The evaluation process involves multiple steps:

CancerHotspots: We check if the variant is present in Cancerhotspots.org^2,3. If so, PM1 is applied.

COSMIC Identifier and PubMed References: We examine the COSMIC (Catalogue Of Somatic Mutations In Cancer)⁴ database for the variant’s identifier. If the variant is associated with more than 20 PubMed references, PM1 is applied.

PM2, BA1, and BS1

We use data from the gnomAD database⁵ to evaluate genetic variants according to the ACMG criteria PM2, BA1, and BS1, which relate to the variant's frequency in the general population:

PM2: If a variant is not present in the gnomAD database, indicating it is rare or novel, we apply PM2.

BA1: If a variant's maximum allele frequency (MAF) in the gnomAD database exceeds 1%, suggesting it is too common to be pathogenic, we apply BA1.

BS1: If a variant's MAF is between 0.3% and 1%, indicating a higher likelihood of being benign, we apply BS1.

PP5/BP6

We search for a corresponding ClinVar entry, and if that entry has been reviewed by an expert panel as either benign/likely benign/pathogenic/likely pathogenic, we apply this criteria.