Codon-level co-occurrences of germline variants and somatic mutations in cancer are rare but often lead to incorrect variant annotation and underestimated impact prediction

Abstract

Cancer cells explore a broad mutational landscape, bringing the possibility that tumor-specific somatic mutations could fall in the same codons as germline SNVs and leverage their presence to produce substitutions with a larger impact on protein function. While multiple, temporally consecutive mutations to the same codon have in the past been detected in the germline, this phenomenon has not yet been explored in the context of germline-somatic variant co-occurrences during cancer development. We examined germline context at somatic mutation sites for 1395 patients across four cancer cohorts (breast, skin, colon, and head and neck) and found 392 codon-level co-occurrences between germline and somatic variants, including over a dozen in well-known cancer genes. We found that for the majority of these co-occurrence events, traditional somatic calling led to an inaccurate representation of the protein site and a significantly lower predicted impact on protein fitness. We conclude that these events often lead to imprecise annotation of somatic variants but do not appear to be a frequent source of driver events during cancer development.

neoepiscopeimproves neoepitope prediction with multivariant phasing

“…However, a limitation of most neoepitope prediction tools is that they consider individual somatic variants in the sequence context of a single reference genome, neglecting interactions between somatic and potentially neighboring germline variants given the widespread genetic variability among individuals [9] . Indeed, somatic and germline mutations may co-occur even within the 3 nucleotide span of a single codon [10] . Such immediate co-occurrences are not uncommon, affecting 17% of cancer patients on average, and can significantly alter variant effects including the amino acid sequence of any predicted neoepitope.…”