SysGen seminar – Agus Salim – 26th August, 2016

darwintree_blue_tips_sysgen2

Assessing Batch Effects and Failed Conversion Rates in Whole-Genome Bisulfite Sequencing Data

Dr Agus Salim
Mathematics & Statistics, La Trobe University

Friday 26th August
11:00am-12:00pm (Followed by light lunch in Level 1 Tearoom, Building 184)
ESJ King Theatre, Medical Building, The University of Melbourne

Abstract
Using simulated and real datasets we investigated the extent and impact of batch effects and failed conversion rates in Whole-Genome Bisulfite Sequencing (WGBS) data. Motivated by ComBat algorithm, we propose a procedure to remove the batch effects and adjust for failed conversion rates, and compare our approach to two popular methods for analysing WGBS data. Simulation studies show that when batch effects are present, removing them increases sensitivity for detecting differentially-methylated loci. Evidence from real dataset shows that batch effects in WGBS data are mostly due to between-flowcell variations. To minimize batch effects, researchers should process biological replicates on the same flowcell.Genetic determinants of rheumatic heart disease in Indigenous Australians

Bio
Dr Agus Salim is a Senior Lecturer at the Department of Mathematic and Statistics, La Trobe University. He came to La Trobe in April 2013 having previously held positions at the Australian National University and National University of Singapore. Agus has over 12 years as a biostatistician and has extensive records of collaboration with clinicians and epidemiologists and has received competitive funding from the National Health and Medical Research Council (NHMRC) for his applied and methodological works. His main methodological interest is in the area of survival analysis with a focus on analysis of nested case-control data and the analysis of high-throughput data from next-generation sequencing.

 

Enquiries: Andrew Siebel (asiebel@unimelb.edu.au)