Phasing or purging: tackling the genome assembly of a highly heterozygous animal species in the era of high-accuracy long reads

Abstract

The revolution of high-accuracy long reads offers unprecedented quality and contiguity in genome assembly. Pacific Biosciences (PacBio) and Oxford Nanopore Technologies have made significant strides in improving their sequencing technologies, yielding reads with error rates below 1% and lengths ranging from kilobases to megabases. These advancements have prompted the development of assembly tools tailored to leverage the enhanced accuracy of long reads. However, the challenge of collapsing haplotypes into high-quality haploid assemblies persists, especially for highly heterozygous genomes. This raises questions about the feasibility and desirability of phased assemblies versus collapsed haploid assemblies. To address these challenges, we benchmarked five assembly tools on ultra-low input PacBio HiFi and Nanopore R10.4 reads from the parthenogenetic nematode species Plectus sambesii. We propose a comprehensive methodology for assessing phased assemblies, repurposing existing evaluation programs to collect haplotype-relevant statistics. Our evaluation criteria include assembly size, contiguity, and completeness, with a focus on assessing the accuracy of phased assemblies by examining duplicated BUSCO orthologs and k-mer spectra. Additionally, we present strategies for generating collapsed assemblies by purging haplotigs. This study provides valuable insights and guidelines for generating high-quality phased and collapsed de novo genome assemblies from highly accurate long reads, particularly beneficial for non-model species genome assembly projects.