Fast linear mixed model computations for GWAS with longitudinal data

Speaker: Emmanuel Lesaffre

Affiliation: KU Leuven, Belgium

Time: Monday 25/02/2013 from 14:30 to 15:00

Venue: Access Grid UWS. Presented from Parramatta (EB.1.32), accessible from Campbelltown (26.1.50) and Penrith (Y239).

Abstract:

Recent genome-wide association studies are directed to identify single nucleotide polymorphisms (SNPs) associated with longitudinally measured traits. In our motivating data set, the bone mineral density (BMD) of more than 5000 elderly individuals was measured at 4 occasions over a period of 12 years. We are interested in SNPs that influence the change of BMD over time. This could be done by fitting a linear mixed model with covariates age, gender, etc but also including each of the SNPs at a time. However, fitting 2.5 million of linear mixed models (1 model per SNP) on a single desktop would take more than a month.

Dealing with such prohibitively large computational time, it is desirable to develop a fast technique. We explored a variety of fast computational procedures. The best approximating procedure is based on a conditional two-step (CTS) approach. This approach approximates the P-value for the SNP-time interaction term from the linear mixed model analysis. Our method is based on the concept of a conditional linear mixed model proposed by Verbeke et al. (2001). A simulation study shows that this method has the highest accuracy of all considered approximations. In fact the P-value obtained by our approach is a close approximation of the P-value obtained by fitting a classical linear mixed model. Applying the CTS approach reduced the computational time needed to analyze the BMD data to 5 hours.

We have explored the robustness of the CTS against different simulation parameters such as sample size, number of measurements, variance-covariance parameters etc. We have also explored the performance of the CTS in case of more complicated residual errors structure (autocorrelation, heteroscedasticity).

Biography: Emmanuel Lesaffre is a professor and Chair, Biostatistics Department, Erasmus University Medical Centre, the Netherlands, and Professor of Statistics at Catholic University of Leuven and Hasselt-Belgium. He is currently visiting the School of Computing, Engineering and Mathematics, UWS. He has more than 300
published papers in applied statistics, more than 25 PhD completions and 15 current. He was the editor of International Statistical Modelling Journal and Associate Editor of Biometrics and many other international journals. He has been the founding chair of the Statistical Modelling Society and President of the International Society for Clinical Biostatistics.