Complete Diploid Assembly of the HG002 Genome

Scientists assembled both parental versions of one person’s genome to test the accuracy of DNA sequencing

On August 6, Cell published a study of HG002, a DNA sample from one person that laboratories use to evaluate sequencing technologies. The authors assembled the maternal and paternal chromosome sets separately. They found no errors in 99.4% of the diploid HG002 genome.

A sequencer reads a genome in fragments, and software places them against a reference sequence, which is a previously assembled map of DNA. In regions that contain repeats or vary substantially between people, a fragment may match several locations. The software can then confuse gene copies, omit a region, or reconstruct its structure incorrectly.

HG002 now provides a benchmark for this type of evaluation, meaning a sequence that serves as a standard for comparison. The new assembly adds 701.4 million DNA bases from the non-sex chromosomes and another 216.8 million bases from both sex chromosomes. This expands the benchmark by 15.3% of the genome. Many of the added regions contain repeats and segmental duplications, which are fragments that the genome has copied and placed at similar locations.

In 2022, the Telomere-to-Telomere research consortium completed the final 8% of the human genome in the CHM13 assembly, which represents a complete sequence of one chromosome set. For HG002, the researchers kept the two inherited versions of the genome from the same sample separate.

Genome Quality Checker compares the benchmark with new sequencing reads, completed assemblies, and lists of DNA variants in which the parental copy carrying each change is known. This comparison shows whether a method retained both copies, distinguished them correctly, and reconstructed repeated regions accurately. The evaluation therefore covers the sample’s own sequence, including regions that are difficult to compare reliably against a general reference genome.

Study coauthor Justin Zook of the United States National Institute of Standards and Technology explains the purpose of the benchmark as follows:

“This result gives technology developers a standard they can use to measure and improve accuracy in the most difficult regions of the human genome.”