Search

Clinical implementation of new prediction models requires evaluation of their validity and utility in a broad range of intended use populations. Here we outline the development and validation of multiple ancestry-specific Polygenic Risk Scores (PRSs) for Coronary Artery Disease (CAD) using a dataset comprising 29,389 individuals from multiple cohorts and diverse genetic ancestry groups. We leverage summary statistics from multiple genome-wide association studies comprising over 850,000 individuals to develop calibrated CAD PRSs with an average Odds Ratio per Standard Deviation (ORxSD) of 1.57 (SD = 0.14). Relative to competing scores, across major genetic ancestry groups these PRSs identify between 26 and 184 additional high risk individuals for every 1,000 people screened. We infer ancestry- specific high risk PRS thresholds and apply these to independent test datasets to identify between 12% and 24% of individuals who are at greater than twice the polygenic risk of CAD compared to the rest of the population. Using these PRSs to reclassify borderline or intermediate 10 year Atherosclerotic Cardiovascular Disease (ASCVD) risk in a cohort of 9,691 individuals improved the classification of those at increased risk of both CAD (Net Reclassification Improvement (NRI) = 13.14% (95%CI 9.23-17.06%)) and ASCVD (NRI = 10.70 (95%CI 7.35-14.05)). Our analyses demonstrate that using PRSs as Risk Enhancers can improve clinical 10 year ASCVD risk assessments and provide an approach for utilizing polygenic information to guide ASCVD prevention efforts.

Gene-flow from an ancestrally differentiated group has been shown to be a powerful source of selectively advantageous variants. To understand whether recent gene-flow may have contributed to adaptation among humans in sub-Saharan Africa, we applied a novel method to identify deviations in ancestry inferred from genome-wide data in 48 populations. Among the signals of ancestry deviation that we find in the Fula, an historically pastoralist ethnic group from the Gambia, are the region that encodes the lactose persistence phenotype, LCT/MCM6, which has the highest proportion of Eurasian ancestry in the genome. The region with the lowest proportion of non-African ancestry is across DARC, which encodes the Duffy null phenotype and is protective for Plasmodium vivax malaria. In the Jola from the Gambia and a Khoesan speaking group from Namibia we find multiple regions with inferred ancestry deviation including the Major Histocompatibility Complex. Our analysis shows the potential for adaptive gene-flow in recent human history.

In the construction of a part-of-speech annotated corpus, we are constrained by a fixed budget. A fully annotated corpus is required, but we can afford to label only a subset. We train a Maximum Entropy Markov Model tagger from a labeled subset and automatically tag the remainder. This paper addresses the question of where to focus our manual tagging efforts in order to deliver an annotation of highest quality. In this context, we find that active learning is always helpful. We focus on Query by Uncertainty (QBU) and Query by Committee (QBC) and report on experiments with several baselines and new variations of QBC and QBU, inspired by weaknesses particular to their use in this application. Experiments on English prose and poetry test these approaches and evaluate their robustness. The results allow us to make recommendations for both types of text and raise questions that will lead to further inquiry.