Approaching an Error-Free Diploid Human Genome

SUMMARYAchieving an error-free diploid human genome remains challenging. We report T2T-YAO v2.0, a telomere-to-telomere complete assembly of a Han Chinese individual, polished to near-perfect base-level and structural accuracy. To systematically identify assembly errors, we developed Sufficient Alignment Support (SAS), an automatic method that flags structural and base-level errors in windows lacking sufficient read support. Building on this, we established a "structural-error-first" polishing strategy, correcting misassemblies using ultra-long ONT reads, followed by base-level refinement with PWC (Platform-integrated Window Consensus). Using these approaches, we resolved all detectable structure and non-homopolymer-related errors outside ribosomal DNA (rDNA) regions. The resulting assembly contains no unsupported 21-mers across sequencing platforms, meeting k-mer-based criteria for an error-free genome. T2T-YAO v2.0 delivers the most perfect East Asian reference to date, with limited issues confined to rDNA arrays and homopolymer tracks, enabling precise genome annotation, benchmarking, and variant discovery--foundation for human genomics and precision medicine.