Share this post

A recent Science article, "Phenotype risk scores identify patients with unrecognized Mendelian disease patterns", from Vanderbilt University Medical Center leverages the phenotypic patterns of Mendelian diseases to detect rare-variant associations and finds that most of the individuals with statistically significant variants were undiagnosed with the target Mendelian disease. The team lead by Joshua Denney used the phenotype data in electronic health records (EHR) at Vanderbilt and mapped HPO terms to these clinical features in the EHR (phecodes – billing codes from the EHR). They computed a “phenotype risk score” (PheRS) that expresses how closely an individual’s symptoms overlap with a Mendelian disease. This score is similar to the genetic risk score approach for analyzing multiple variants against a single phenotype.

Science-16-March-2018-6381.cover-sourceUsing a cohort of 21,701 adults of European ancestry genotyped on the Exome BeadChip, they performed an association analysis. They computed the PheRS for 1024 Mendelian diseases for which they had sufficient genotype data. They looked for the association between the PheRS and 6188 rare variants and found 18 significant associations, with four genes having a dominant mode of inheritance. Of the 13 recessive genes, ClinVar and HGMD provided evidence of pathogenicity for seven of the genes. Only eight of the 807 individuals with variants in these 18 genes had been clinically diagnosed suggesting that there may be many in the population with undiagnosed Mendelian disease.

The team continued with analyses by testing for additional rare variants segregating with high-PheRS individuals. Via whole-exome sequencing they found that some individuals had an additional rare non-synonymous variant in the target gene; pathogenicity was confirmed via in vitro studies. This technique can also be applied to variant interpretation; using ACMG guidelines, 10 of the variants were interpreted as VUS. By adding the PheRS and results from in vitro studies, four of these variants could be converted to "pathogenic” or “likely pathogenic”. This new method employed by the team is easy to implement and can be applied to other large populations and could facilitate the discovery of pathogenic variants and also aid in interpretation.

Read the full Science article.