Approaching an Error-Free Diploid Human Genome

Achieving a truly error-free human diploid genome remains a central challenge in genomics. Here, we present T2T-YAO v2.0, a nearly perfect diploid genome assembly for a Han Chinese individual. We first developed Sufficient Alignment Support (SAS), a sensitive alignment-based metric that identifies both structural and base-level assembly erroneous windows by quantifying inadequate read support across sequencing platforms. Unlike conventional methods that detect erroneous k-mers ( e.g. , Merqury) or inconsistencies between assemblies and aligned reads from single platform, SAS integrates supportive information from various platform and avoids false positives caused by sequencing bias and artificial alignments. Employing a “Structural Error-first” polishing strategy, prioritizing structural error correction using ultra-long Oxford Nanopore reads, we iteratively refined the assembly with high-accuracy PacBio and Element sequencing data. This approach eliminates all detectable structural and non-homopolymer-related base-level errors outside ribosomal DNA clusters. The final T2T-YAO v2.0 assembly exhibits complete non-split ONT read coverage and free of erroneous 21-mers outside sequencing data, representing an error-free human genome by Merqury standards. Our results not only highlight the limitations of current genome quality assessment methods but also provide a roadmap for constructing high-fidelity diploid references essential for precision medicine and genome-based diagnostics.