Abstract
Haplotype phasing, the process of determining which genetic variants are physically located on the same chromosome, is crucial for genetic analyses. Here, we benchmark SHAPEIT and Beagle, two state-of-the-art phasing methods, on two large datasets: >8 million research-consented 23andMe, Inc. customers and the UK Biobank (UKB). Remarkably, both methods' median switch error rate (SER) (after excluding single SNP switches, which we call "blips") is 0.00% across all tested 23andMe trio children and 0.026% in British samples from UKB. Across UKB samples, switch errors predominantly occur in regions lacking identity-by-descent (IBD) coverage. SHAPEIT and Beagle excel at intra-chromosomal phasing, but lack the ability to phase across chromosomes, motivating us to develop HAPTiC (HAPlotype Tiling and Clustering), an inter-chromosomal phasing method that assigns paternal and maternal variants genome-wide. Our approach uses IBD segments to phase blocks of variants on different chromosomes. HAPTiC represents the segments a focal individual shares with their relatives as nodes in a signed graph and performs spectral clustering. We test HAPTiC on 1,022 UKB trios, yielding a median per-site phase error of 0.13% in regions covered by IBD segments (45.1% of sites). We also ran HAPTiC in the 23andMe database and found a median phase error rate of 0.49% in Europeans (100% of sites) and 0.16% in admixed Africans (99.8% of sites). HAPTiC enables analyses that require the parent-of-origin of variants, such as association studies and ancestry inference of untyped parents.