The Effects of True Match Rate, Imputation Accuracy, and Population Structure on a Method for Genetic Record Matching

Abstract

As the number and size of genetic databases continues to expand, linking between them will become increasingly important, especially in forensic contexts. Edge et al. have proposed a best-in-class method for matching profiles between an STR sequenced and a SNP sequenced database. Through the use of additional genetic profiles and newly available whole genome sequencing data, we assess how this method is affected by the true match rate between the two databases, imputation accuracy, and population structure. We find that matching accuracy decreases monotonically as the true match rate decreases. Greater imputation accuracy from whole genome sequenced profiles enables substantially higher matching accuracy. Finally, accounting for population structure can modestly increase matching accuracy, but must be done carefully.