Examination Committee

Prof Wai Ho MOW, ECE/HKUST (Chairperson)
Prof Weichuan YU, ECE/HKUST (Thesis Supervisor)
Prof Wei ZHANG, ECE/HKUST
 

Abstract

In genome-wide association studies (GWAS), detecting the interactions among single nucleotide polymorphism (SNP) pairs and phenotypes is important to reveal the relations between genotypes and genetic diseases. Recently, a Boolean operation-based screening and testing (BOOST) method was proposed to detect interactions with log- linear models. As the interaction detection is quite parallel, a GPU-based implementation of BOOST method, named GBOOST, was made available for acceleration. Both BOOST and GBOOST method didn’t take covariates into consideration in the models, which may lead to in inaccurate or even wrong interaction results under some circumstance.
 
In the thesis, the covariate adjusted interaction detection tool (BOOST 2.0/GBOOST 2.0) will be presented. BOOST 2.0 is a CPU multi-threaded version of the advanced method and GBOOST 2.0 is a GPU-based implementation. We will introduce the log-linear models and the solutions to maximum log-likelihood of models used in the method. Then the CPU multi-threaded and GPU implementation will be illustrated.
 
The performance comparison of BOOST/GBOOST with BOOST 2.0/GBOOST 2.0 will be presented by experiments on simulated data. We will also show the discoveries we found in real data with BOOST 2.0 and GBOOST 2.0.