Abstract
Background Polygenic scores provide an indication of an individual’s genetic propensity for a trait within a test population. These scores are calculated using results from genetic analysis conducted in discovery populations. Where the test and discovery populations have similar ancestries, the predictions are better than when the ancestries differ. As many of the genetic analyses are conducted in European populations this hinders the potential for maximising predictions in many of the currently underrepresented populations in research. Methods To address this, UP and Downstream Genetic scoring (UPDOG) was developed to consider the concordance of genetic variation around lead variants between the discovery and test cohorts before calculating polygenic scores. Where there was non-concordance between the discovery cohort and an individual in the test cohort, the lead variant’s effect was down weighted for that individual. Results UPDOG was tested across four ancestries and six phenotypes and benchmarked against five existing tools for polygenic scoring. In approximately two-thirds of cases UPDOG improved trans-ancestral prediction, although the increases were small. Conclusions The development of novel methodologies aimed at maximising the efficacy of polygenic scores for the global population is of high importance and enables progress towards personalised medicine and universal equality in healthcare.