Identifying Mendelian disease genes with the variant effect scoring tool

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Whole exome sequencing generates numerous rare variants of unknown significance.
• Efficient computational tools are crucial for identifying disease-causing genetic variants and genes.

Purpose of the Study:

• To develop and evaluate the Variant Effect Scoring Tool (VEST) for prioritizing rare missense variants in human disease.
• To assess the utility of aggregating VEST scores for identifying Mendelian disease genes.

Main Methods:

• VEST is a machine learning classifier trained on disease mutations and neutral variants.
• Performance was benchmarked against existing methods like PolyPhen2 and SIFT.
• Gene-level scores were derived by aggregating variant p-values across multiple exomes.

Main Results:

• VEST demonstrated superior performance (ROC AUC = 0.91) compared to other methods.
• Aggregate VEST gene scores successfully ranked known causal genes (DHODH, MYH3) for Mendelian disorders.
• The tool effectively identified candidate disease genes in case studies.

Conclusions:

• Aggregating bioinformatics variant scores into gene-level scores enhances disease gene discovery.
• VEST provides a valuable bioinformatics resource for large-scale exome sequencing studies.
• VEST is accessible as a software package and via the CRAVAT web server.