MRC INTEGRATIVE EPIDEMIOLOGY UNIT (IEU) 

SEMINAR SERIES

 Tuesday, 25th October, 2016

16.00 – 17.00 – Room OS6, Second Floor, Oakfield House

Peter Newcombe
Senior Investigator in the Statistical Genomics Group
MRC Biostatistics Unit, Cambridge

"JAM: A scalable Bayesian framework for joint re-analysis of pubished one-at-a-time SNP effects"

 

Abstract

Recently, large scale genome wide meta-analyses - accumulating information over tens of thousands of people - have boosted the number of known genetic risk factors for some traits into the tens and hundreds. However, the availability of many correlated single nucleotide polymorphisms (SNPs) presents analytical challenges and typically variants are only analysed one-at-a-time. This complicates the ability of fine-mapping to identify a small set of SNPs for further functional follow up.

I will describe “JAM” - a new and scalable algorithm for the re-analysis of published marginal associations under joint multi-SNP models, in which correlation is accounted for according to estimates from a reference dataset. SNPs which best explain the joint pattern of effects are highlighted via an integrated Bayesian penalised regression framework. Through a realistic simulation study, including an application to 10,000 SNPs, we demonstrate substantial gains in the proportion of true signals among top ranked SNPs (positive predictive value) using our multivariate framework.

I will go on to present a real data application to published results from MAGIC (Meta-Analysis of Glucose and Insulin Related Traits Consortium) - a GWAS meta-analysis of more than 15,000 people, in which we re-analyse several top loci associated with glucose levels two hours after oral stimulation. Our algorithm was able to rule out many SNPs as false positives and for one gene, ADCY5, joint modelling of the pattern of effects across the locus highlighted an alternative, and more plausible, SNP to the reported index.

Biography

Peter Newcombe has worked as a senior investigator in the statistical genomics group at the MRC Biostatistics Unit, Cambridge, since joining in 2012. Previously I did a Ph.D. in statistical genetics at the London School of Hygiene & Tropical Medicine and, subsequently, spent three years in the statistical genetics group at GlaxoSmithKline. My current research is focussed on problems relating to summary statistics, Mendelian randomisation, and genomic prediction.

 

ALL WELCOME