蜜桃传媒app

Low-dimensional representations of high-dimensional biobank data

Alex Diaz-Papkovich, Brown University
Tuesday January 14, 12-1pm
Zoom Link:听
In Person: 550 Sherbrooke, Room 189

Abstract:听With cheap and efficient DNA sequencing readily available, scientists have access to the genetic data of millions of individuals through biobanks. Statistical analyses of these data can give important insights into medical genetics, demographic history, and population health. However, biobanks are as complex as they are rich鈥攃omplications from sampling methodology, confounding with environmental factors, and idiosyncratic population histories ensure that analysis is no easy task. Methods like dimensionality reduction, clustering, and visualization can bring to light patterns in data that would otherwise go unnoticed. In this seminar, I will give an outline of how manifold learning and density clustering can uncover signal in multiple biobanks, highlighting patterns between genetics, geography, the environment, and health measures. I will also discuss the role low-dimensional representations can play in illustrating biobank composition and human genetic diversity, as well as their ethical and social implications.