The revised diploid genome sequence of an individual human: A workflow for near telomere-to-telomere assemblies using only Oxford Nanopore Technologies Simplex-reads

The first draft diploid genome assembly (HuRef) of an individual in 2007 was a milestone in genomics. Here, we report HuRef2.0, a revision of HuRef assembled using only Oxford Nanopore Technologies (ONT) Simplex-reads and the hifiasm assembler. Results are close in continuity to the telomere-to-telomere (T2T) assemblies but were assembled from standard DNA samples without using multiple sequencing and mapping technologies, including ultra-long-reads and/or proximity-ligation. Three ONT flowcells (~103× coverage) from fresh blood DNA produced an assembly comprising 26 contigs, with gapless assembly of 23 chromosomes (Chr). Two gaps on ChrY were locally assembled to yield the final T2T-assembly, HuRef2.0, with base accuracy >Q60.34 and 2,393 phase blocks with an NG50 value of 2.36 Mb. Assembly from a single ONT flowcell (~35× coverage) consistently produced an assembly more contiguous than GRCh38.p14, providing a foundation for further optimization and scaling. Assembly quality was assessed by direct chromosome-level alignments to a reference genome, variant calling, and annotation of gene-rich regions at Chr22q11, the extended MHC locus, and at several difficult to assemble regions, including the rRNA gene clusters, the subtelomeric region on Chr4q35, ChrY, and the centromeres. More accurate but shorter Pacific Biosciences HiFi-reads produced less contiguous assemblies than from equivalent coverage of error-corrected ONT reads, indicating the importance of read-length. We compared HuRef2.0 to an assembly of a low-passage EBV-transformed lymphoblastoid cell line derived from the same donor. At least at a 400 kb resolution, we did not observe notable structural differences in the two assemblies.