Quantitative Issues in Cancer Research Working Seminar

Heng Li, Ph.D.
Assistant Professor
Biomedical Informatics, Harvard Medical School
Biostatistics and Computational Biology, Dana-Farber Cancer Institute

“The Assembly of Human Pangenome”

ABSTRACT: The primary GRCh38 assembly represents the genome of a single haplotype and misses thousands of structural variations in each individual. To encode and study these complex variations, we developed hifiasm and minigraph to assemble PacBio high-fidelity reads into long haplotype contigs and then to represent these contigs with a pangenome graph. Applying the tools to samples from the Human Pangenome Project, we can produce a human pangenome graph composed from 58 haplotype assemblies with their contig N50 reaching 55Mb. In this talk, we will explain the rationale behind the hifiasm assembly algorithm, the minigraph graph construction algorithm and showcase complex variations we can learn from a pangenome graph.