Calling germline and mosaic variants from long genomic and RNA-seq reads

Dana-Farber Cancer Inst, Boston MA

Investigators

Abstract

PROJECT SUMMARY The rapidly reduced cost and increased base quality of long sequence reads enable the accurate calling of germline and mosaic sequence variations thanks to phasing and the power to resolve repetitive regions. Although multiple callers have been developed for long-read variant calling, tools are still missing for several important applications such as variant calling from long RNA-seq reads, mosaic small variant calling for long reads and accurate mosaic structural variant calling. In addition, existing tools often target one type of variants and treat small variants, structural variants and phasing as separate problems. This reduces the power of long reads and makes the current tools difficult to use for biologists. Here, we plan to address these issues with three proposals: (1) developing a variant caller for calling germline variants, mosaic mutations and RNA editing events from long RNA-seq reads; (2) improving the accuracy of mosaic and somatic structural variant calling using pangenome graphs and the de novo assembly of the normal bulk sample; (3) jointly calling germline/mosaic small/structural variants from long genomic reads by using integrated phasing and local realignment or reassembly methods. Upon completion, this proposal will result in new computational tools for tasks not achievable with current methods and for the calling of most types of variants to higher accuracy.