Rapid and accurate genotype imputation from low coverage short read, long read, and cell free DNA sequence

Abstract

Inexpensive and accurate genotyping methods are essential to modern genomics and health risk prediction. Here we introduce QUILT2, a scalable method for genotype imputation using low-coverage sequencing. QUILT2 contains two technical innovations compared to our previous method QUILT, which enable rapid imputation from haplotypes derived from biobank scale whole genome sequenced data. Further, QUILT2 contains a methodological innovation that enables imputation from the 3 haplotypes present in cell free non-invasive prenatal testing (NIPT) data. Through comprehensive benchmarking, we show that QUILT2 maintains the accuracy of QUILT across diverse sequencing data (e.g. ONT long reads, ancient DNA), but is much faster and more memory efficient. In addition, we show that accurate imputation using NIPT enables accurate GWAS and PRS for both mother and fetus. This creates both clinical possibilities, and, if phenotypes can be collected alongside clinical NIPT, the potential to enable future large GWAS.