Computational Tools and Analyses for Improved Inference of Variant Effects

Personalized medicine requires rapid and accurate classification of pathogenic human variation. With over 50% of clinically interpreted missense variants classified as “variants of uncertain significance” (VUSes), multiple approaches are needed to improve variant interpretation. Multiplexed assays of variant effect (MAVEs) can experimentally test nearly all possible missense variants in selected protein targets, while computational methods seek to infer missense variant impacts using statistical modelling. Here I describe work to assist and exploit both MAVE and computational studies.To assist in planning and to promote collaboration and efficient communication for MAVE studies, I developed: 1) strategies to prioritize genes likely to have a greater impact on clinical variant interpretation, 2) MaveQuest, a resource to help researchers identify MAVE target genes and explore potential assays, and 3) MaveRegistry, a community resource for sharing MAVE progress and finding collaborators. To compare the performance of variant effect predictors and to exploit both experimental and computational information about variant impact, I 1) assessed computational variant effect predictors using a large prospective cohort, and 2) developed a pipeline to screen human pseudogenes for which different genetic variant interpretation might re-classify these pseudogenes to be protein-coding genes, refining human genome annotation.

Department dc:contributor.department

Molecular Genetics

Year dc:date.issued

2022

Author dc:creator

• Kuang, Da

Advisor dc:contributor.advisor

• Roth, Frederick P

Statement dc:rights

• Attribution 4.0 International

Licence dc:rights.uri

• http://creativecommons.org/licenses/by/4.0/

Handle dc:identifier.uri

http://hdl.handle.net/1807/123134

OAI identifier oai:identifier

oai:utoronto.scholaris.ca:1807/123134