Association Study Of GRIK4 Gene With Schizophrenia And Improvement Of EM Algorithm For Haplotype Inference

Schizophrenia is one of the most severe mental disorders, and it is characterized by psychotic symptoms and by cognitive, affective, and psychosocial impairment, with a worldwide incidence of 1%. It is a highly heritable and complex disease, in a polygenic, non-Mendelian fashion, possibly caused by the interaction of numerous minor effect genes and the influence of the environment. Many hypotheses were proposed, but none of them can explain all the etiologies of schizophrenia. The molecular mechanism of it remains elusive. Linkage analysis and association study are main genetics methods for detecting candidate genes for schizophrenia.Glutamate is a major excitatory neurotransmitter of the mammalian central nervous system. A large number of studies have focused on dysfunctions in the glutamatergic pathway as a major susceptibility factor for schizophrenia. Glutamate receptors are widely expressed in the central nervous system and play a fundamental role in synaptic plasticity and in all the processes underlying learning and memory. In this current thesis, I will report association study between the kainite type glutamate receptor gene, GRIK4, and schizophrenia. The GRIK4 gene encodes the high affinity subunit (KA1), and it is located on chromosome 11q22.3.We examined five SNPs within GRIK4 in our samples and performed linkage disequilibrium study. Three of the five SNPs were reported associated with schizophrenia in a Scottish population. 576 Chinese Han samples were used including 288 schizophrenia patients and 288 controls in a case-control study. No significant difference was observed in the analysis of either single nucleotide polymorphisms or haplotype-based analysis.In molecular genetics research, haplotype phase information plays an important role in detecting genetic variants that increase susceptibility to human diseases. However, haplotypes cannot easily be acquired. Molecular haplotyping methods are labor-intensive, low-throughput, and very expensive. The application of statistical methods to infer the haplotype phase in samples of diploid sequences is a very effective and cost-efficient method. Several computational and statistical methods have been developed for haplotype inference, including Clark's algorithm, EM algorithm, Bayesian method and so on. Because of its interpretability and stability, the EM algorithm becomes the most popular one. However, it can not handle too many loci and multiallelic loci, because of the memory constraint. To overcome this deficiency, we developed PL-CSEM program, which combined the Partition Ligation and Combination Subdivision strategies with the standard EM algorithm. Large numbers of data sets were used to evaluate the performance of the program, and the results show that our program not only can handle multiallelic loci nicely, but also perform better than Qin et al's PLEM program for many linked SNPs loci.