Jianjun Liu from the Genome Institute of Singapore presented at the Nanopore Community Meeting 2023 in Singapore a session entitled “Redefining telomere-to-telomere genome assembly strategy using the Oxford Nanopore platform.” They did a study in collaboration with Oxford Nanopore Technologies (ONT) to assemble a telomere-to-telomere (T2T) human genome with only Nanopore reads. Liu spoke about primary assemblies with a consensus sequence constructed from fragmented DNA sequencing data. Now there are tools to resolve haplotype assembly of genomes. Liu noted that in 2003 the first “essentially complete” human genome was published. In 2022, the first complete haploid T2T human genome was published. Telomere-to-telomere haplotype resolved assembly requires highly accurate long reads, ultra long reads >200 kb), HiC/PoreC and an algorithm. To build the graph based on overlapping information of reads, Liu explained that raw reads need to be cleaned. Liu shared a table with current trends of data utilization for T2T assembly. Most approaches use hybrid assemblies using both short and long reads. Liu’s team generated data from samples from the Asian Reference Genome project from a trio. The assembly was presented using different tools: Verkko and Hifiasm. The results were very different. Hifiasm produced slightly longer reads and was much faster. Assembly results were also presented in a chromosome-level graph for maternal and paternal classifications. Both Verkko and Hifiasm covered most of the chromosome with some regions near centromeres lacking coverage. The phasing accuracy was dependent on parental information. Coverage saturation was also assessed, and Liu noted that fewer PoreC reads than expected were needed. From this study, Liu concluded that “constructing haplotype-resolved high-quality telomere-to-telomere genomes is achievable exclusively through ONT datasets, including high-quality duplex, ULONT, and PoreC.” Both HIfiasm and Verkko can be used to generate T2T assemblies with only ONT reads, and Hifiasm may be the best option for population-scale samples and phasing. This session was the first that I have watched describing comparisons on ONT only T2T human genomes. I appreciate the comparison of tools and their benefits.
