Short communication: imputation accuracy of host genomic data from metagenomic sequence information

Metagenomic sequencing is the process of extracting all the genomic information from a given sample. Most metagenomic studies remove any host reads as a matter of course. However, host reads can be used as the basis for genotype imputation to obtain whole genomic sequences. The accuracy of these imputed genotypic calls from a bovine ocular sample was determined by comparing results to those from a commercial array. Overall, imputed genotype calls proved to have a high concordance with array genotype calls (average concordance of 83% and correlation of 0.81 with no filtering). Accuracy increased as filters for host read depth and imputed call confidence were implemented. With filters in place, the average percent concordance was 98% (88% to 99%) while the mean correlation was 0.98 (0.89 to 0.99). Further, identity verification of the metagenomic samples can be carried out if the host is genotyped on another platform.