An accurate phasing tool via high-quality heterozygous variations

Abstract

Long-read phasing has become a widely adopted approach for reconstructing paternal and maternal haplotypes in diploid genomes. However, most existing phasing methods treat all variants equally, even though low-quality variants frequently introduce switch errors. We present a new phasing method, snpphaser, which first phases high-quality variants and then leverages them to filter out low-quality variants that are inconsistent with the high-confidence haplotypes. We benchmarked snpphaser against state-of-the-art phasing methods across nine diverse datasets. The results demonstrate that snpphaser achieves superior contiguity and accuracy compared to existing approaches.