Title | Two-stage designs to identify the effects of SNP combinations on complex diseases |
Authors | Kang, Guolian Yue, Weihua Zhang, Jifeng Huebner, Marianne Zhang, Handi Ruan, Yan Lu, Tianlan Ling, Yansu Zuo, Yijun Zhang, Dai |
Affiliation | Peking Univ, Inst Mental Hlth, Key Lab Mental Hlth, Minist Hlth, Beijing 100083, Peoples R China. Chinese Acad Sci, Acad Math & Syst Sci, Inst Syst Sci, Key Lab Syst & Control, Beijing 100080, Peoples R China. Michigan State Univ, Dept Stat & Probabil, E Lansing, MI 48824 USA. Peking Univ, Inst Mental Hlth, Key Lab Mental Hlth, Minist Hlth, 51 Hua Yuan Bei Rd, Beijing 100083, Peoples R China. |
Keywords | SNP pattern synergistic block association study MULTIFACTOR-DIMENSIONALITY REDUCTION DETECTING GENE-GENE BREAST-CANCER ASSOCIATION SCHIZOPHRENIA EPISTASIS |
Issue Date | 2008 |
Publisher | journal of human genetics |
Citation | JOURNAL OF HUMAN GENETICS.2008,53,(8),739-746. |
Abstract | The genetic basis of complex diseases is expected to be highly heterogeneous, with many disease genes, where each gene by itself has only a small effect. Based on the nonlinear contributions of disease genes across the genome to complex diseases, we introduce the concept of single nucleotide polymorphism (SNP) synergistic blocks. A two-stage approach is applied to detect the genetic association of synergistic blocks with a disease. In the first stage, synergistic blocks associated with a complex disease are identified by clustering SNP patterns and choosing blocks within a cluster that minimize a diversity criterion. In the second stage, a logistic regression model is given for a synergistic block. Using simulated case-control data, we demonstrate that our method has reasonable power to identify gene-gene interactions. To further evaluate the performance of our method, we apply our method to 17 loci of four candidate genes for paranoid schizophrenia in a Chinese population. Five synergistic blocks are found to be associated with schizophrenia, three of which are negatively associated (odds ratio, OR < 0.3, P < 0.05), while the others are positively associated (OR > 2.0, P < 0.05). The mathematical models of these five synergistic blocks are presented. The results suggest that there may be interactive effects for schizophrenia among variants of the genes neuregulin 1 (NRG1, 8p22-p11), G72 (13q34), the regulator of G-protein signaling-4 (RGS4, 1q21-q22) and frizzled 3 (FZD3, 8p21). Using synergistic blocks, we can reduce the dimensionality in a multi-locus association analysis, and evaluate the sizes of interactive effects among multiple disease genes on complex phenotypes. |
URI | http://hdl.handle.net/20.500.11897/397184 |
ISSN | 1434-5161 |
DOI | 10.1007/s10038-008-0307-x |
Indexed | SCI(E) |
Appears in Collections: | 第六医院 |