A complete diploid human genome benchmark for personalized genomics.

Area of Science:

• Genomics
• Bioinformatics
• Human Genetics

Background:

• Human genome resequencing faces challenges with reference biases, leaving complex genomic regions unmapped.
• Existing variant benchmarks are limited and do not accurately assess these difficult-to-map regions.

Purpose of the Study:

• To develop a comprehensive, highly accurate telomere-to-telomere genome benchmark for the complete diploid HG002 genome.
• To improve the assessment of variant calling and genome assembly methods in previously unmapped genomic regions.

Main Methods:

• Generation of a telomere-to-telomere genome benchmark covering 99.4% of the complete diploid HG002 genome, including previously unmapped autosomal and sex chromosome sequences.
• Development of tools for measuring the accuracy of sequencing reads, phased variant call sets, and genome assemblies against a diploid reference.
• Diploid annotation of genes, transposable elements, segmental duplications, and satellite repeats.

Main Results:

• The benchmark achieves near-perfect accuracy across 99.4% of the HG002 genome, adding 15.3% of sequence absent from prior benchmarks.
• State-of-the-art de novo assembly methods resolve 2-7% more sequence and show an order of magnitude improvement in variant calling accuracy (1 error/100 kb).
• The benchmark includes a diploid annotation of 39,144 protein-coding genes across both haplotypes.

Conclusions:

• The telomere-to-telomere genome benchmark significantly enhances the accuracy and completeness of human genome analysis.
• This benchmark is expected to accelerate the development of cost-effective complete genome sequencing methods.
• Adoption of this benchmark will expand genomic medicine to the entire genome, ushering in an era of personalized genomics.