SysGen Seminar – Michael Beer – 24th January, 2017


More Information

Andrew Siebel

T: +61 3 8344 0707

Michael Beer

Department of Biomedical Engineering and McKusick-Nathans Institute of Genetic Medicine, John Hopkins School of Medicine

Tuesday 24th January
FW Jones Theatre, 3rd Floor, Medical Building, The University of Melbourne

Making Sense of Intergenic Variants with Machine Learning

The vast majority of SNPs associated with common human disease are intergenic and likely regulatory, but linkage disequilibrium makes it difficult to identify functional SNPs without direct experimental validation.  We have made progress understanding these intergenic variants by using epigenomic data to train a classifier whose scoring function encodes the relative regulatory importance of individual sequence features in a specific cell-type.  Mutations induce changes in these feature scores which determine the predicted impact of the mutation, a score which we call deltaSVM (Lee–Beer, Nature Genetics 2015). We show that deltaSVM is about 10x more accurate at predicting dsQTLs than other methods, and we have also used deltaSVM to successfully predict the expression change in massively parallel reporter assays in mouse liver, human cell lines, and mouse retina.  We are currently using deltaSVM to identify and test SNPs associated with a range of GWAS studies, and we present comparisons of deltaSVM to other computational approaches, including PWMs, other kmer-based approaches, and deep neural networks.

Michael A. Beer was a research scientist at the Princeton Plasma Physics Laboratory after receiving his Ph.D. in theoretical plasma physics from Princeton University in 1995.  Following the sequencing of the human genome, he began postdoctoral research in genomics at Princeton University.  He is now an Associate Professor at Johns Hopkins and directs the computational regulatory genomics laboratory in the Dept of Biomedical Engineering and the McKusick-Nathans Institute of Genetic Medicine, where he develops computational methods to understand how gene regulatory information is encoded in genomic DNA, and how variation in regulatory DNA elements contributes to human disease and evolution. He has published over 80 scientific papers and has over 7000 citations.  He was a Searle Scholar and serves on the Editorial Board of Genome Research.

Enquiries: Andrew Siebel (