SRA

Inexpensive genotyping methods are essential to modern genomics. Here we present QUILT, which performs diploid genotype imputation using low-coverage whole genome sequence data. QUILT employs Gibbs sampling to partition reads into sets reflecting maternal and paternal origin, which facilitates rapid haploid imputation using large reference panels. We show this partitioning to be accurate over many megabases, and that this enables highly accurate imputation that near theoretical limits and out-performs existing methods. Moreover, QUILT can impute accurately using diverse sequencing technologies, including using long reads from Oxford Nanopore Technologies, and a novel form of low-cost barcoded short-read (linked-read) Illumina sequencing called haplotagging, with the latter showing improved accuracy particularly at coverages <1X. We show that relative to commonly used DNA genotyping microarrays, QUILT offers improved accuracy at reduced costs, with particularly large improvements for genotyping in diverse populations that are traditionally underserved in modern genomic analyses, with a near doubling of imputation accuracy at rare SNPs. Finally, QUILT can be used for (4-digit) HLA typing, enabling the first accurate typing from low-coverage sequence data.