躁鬱症是常見的一種複雜型精神疾病。它透過體染色體及X染色體在家族的個體中遺傳，除此之外它也受到環境因素影響，如化學要素或情緒壓力等。DNA甲基化在表觀遺傳學中佔重要角色，它在DNA遺傳序列以外建立改變生物性狀並且可以遺傳的機制。於是我們以偵測DNA甲基化的CpG 小島微陣列晶片(包含13056個探針) 來分析躁鬱症病患與正常人精子裡DNA甲基團的分佈，並藉由統計學裡的淨相關來計算微陣列數據中探針間的相關性。無論是質譜儀、基因體染色質免疫沉澱或是其他高通量實驗數據，圖型模型是一個適合資料探勘與整合的分析方式。我們以圖型化高斯模型來表示這些CpG小島中甲基團密度變化量的相依性，並且探討圖型結構中的特性 ─ 模數、度協調、叢集係數等─及註解其生物意義。 Manic depression disorder is a complex heritable disease in that its heredity comes from genetic abnormalities in somatic and X chromosomes. Beside genetic factor, environment factors also affect the disease's susceptibility. DNA methylation plays a role in epigenetics and is considered to serve as a bridge between genetic and environmental factors. In this thesis, we describe a genome-wide DNA methylation dataset of 19 bipolar disorder (manic depression) and 18 normal individuals using human CpG island microarrays (containing 13056 probes). Then we analyze the methylation relations among the measuring probes using Graphical Gaussian models for each phenotypic group, obtaining networks of co-methylation. Modularity, assortativity and clustering coefficient properties of the networks are calculated. Genes involved in co-methylation are grouped and annotated by Gene Ontology terms. Candidate genes have some implications to the disorder.
