MosaicFlye: Resolving Long Mosaic Repeats Using Long Reads

摘要

Although long-read sequencing technologies opened a new era in genome assembly, the problem of resolving unbridged repeats (i.e., repeats that are not spanned by any reads) such as long segmental duplications in the human genome remains largely unsolved, making it a major obstacle towards achieving the goal of complete genome assemblies. Thus, long-read assemblers currently face the same repeat-resolution challenge that short-read assemblers faced a decade ago (albeit at a different scale of repeat lengths). Long error-prone reads and short accurate reads have their strengths and weaknesses, e.g., short reads may resolve some long repeats that are difficult to resolve with long reads. For example, diverged copies of a long repeat (e.g., copies differing by 3%) often don't share k-mers (for a typical k-mer size used in short-read assemblers) and thus are automatically resolved by the de Bruijn graph-based assemblers. In contrast, long-read assemblers face difficulties resolving such repeats since repeat copies with a 3% divergence are difficult to distinguish using the error-prone reads with error rates exceeding 10%. Thus, long-read assemblers trade the ability to resolve the unbridged but divergent repeat copies for the ability to resolve bridged repeats.