Uncalling Structural Variant Artifacts
Structural variants (SVs) have significant genetic consequences but remain challenging to detect accurately. Although many long-read SV callers have been published, they suffer from high error rates in low-complexity regions and often lack the precision required to call rare somatic or de novo SVs. In this talk, I will showcase various artifacts in germline and somatic SV calling and explain the underlying mechanisms behind them. I will then present effective computational strategies, including assembly and pangenome-based approaches, to substantially alleviate these SV calling errors to an accuracy practical for downstream applications.
Heng Li, Ph.D.
Harvard Medical School
Heng Li is an Associate Professor of Biomedical Informatics at Harvard Medical School and Dana-Farber Cancer Institute. He has deep expertise in developing high-performance computational algorithms for large-scale biological sequence analysis. His work has fundamentally shaped the practices in high-throughput sequence data analysis and receives tens of thousands of citations per year. All research labs that rely on sequencing data use formats designed by him; many of them routinely run alignment, assembly, variant calling or pangenome algorithms developed in his group. He was a Sloan fellow and is a distinguished fellow of the International Society of Computational Biology (ISCB). He is also a recipient of the Benjamin Franklin Award for Open Access in Life Science.