A comparative medical genomics approach may facilitate the interpretation of rare missense variation

Purpose: To determine the degree to which likely causal missense variants of single-locus traits in non-human livestock and domestic species have features suggestive of pathogenicity in a human genomic context. Methods: We extracted missense variants from the Online Mendelian Inheritance in Animals database for nine animals (cat, cattle, chicken, dog, goat, horse, pig, rabbit, sheep), mapped coordinates to the human reference genome, and annotated variants using standard genome analysis tools. We also searched a private commercial laboratory database of genetic testing results from >400,000 individuals with suspected rare disorders. Results: Of the 339 variants that were mappable to the human genome and impacted identical residues, 56 had been previously classified with respect to pathogenicity: 31 (55.4%) pathogenic/likely pathogenic, 1 (1.8%) benign/likely benign, and 24 (42.9%) uncertain/other. The odds ratio for a pathogenic/likely pathogenic classification in ClinVar was 7.0 (95% confidence interval: 4.1-12.0, p<0.0001), compared to all other germline missense variants in these same 220 genes. The remaining 283 variants disproportionately had allele frequencies and REVEL scores that supported pathogenicity. Conclusion: Cross-species comparisons could facilitate the interpretation of rare missense variation. These results provide further support for comparative medical genomics approaches that connect big data initiatives in human and veterinary genetics.