上海科技大学知识管理系统

Haplotype-resolved genomes play a pivotal role in understanding evolutionary histories, genetic variations, and complex traits. While trio binning is an effective technique for haplotype phasing when parental information is available, its practical application is often hindered by the absence of parental data.

We propose a novel approach DipHiphaser that integrates PacBio HiFi reads with Hi-C chromatin interaction data to resolve chromosome-level haplotypes without parental data. The core idea of our method is to leverage SNP-mer pairs, defined as pairs of k-mers each with a single nucleotide polymorphism (SNP) located exactly at the middle point, to identify haplotypes. These SNP-mer pairs are initially derived from HiFi reads and largely represent haplotype information as they are mostly mapped to heterozygous regions of the genome. Subsequently, we connect these SNP-mer pairs based on overlaps in HiFi data and further phase and assemble them into chromosome-haplotype blocks using Hi-C data. When applied to human genomes and a bird, DipHiphaser achieves an impressive phasing accuracy of HiFi reads exceeding 97% at both the haplotype and chromosome levels, outperforming trio binning. These phased reads can be independently assembled to achieve high contig phasing accuracy, closely approaching the top-performing haplotype-resolved assembly pipelines. DipHiphaser provides an accurate way of phasing reads at chromosome-level which leads to haplotype-resolved diploid genome assembly without the requirement of parental data and has great potential for studying haplotype variation and inheritance for diploid species.