利用多重策略發掘基因網路

Abstract

受到調控蛋白所調控的基因，其基因的表現程度與產生調控蛋白的基因有著高度的相關性，研究其相關性就系統生物學領域而言是一個相當重要的課題，特別是在基因體及蛋白質體上之整合扮演重要的角色。然而近年來有許多人提出了許多的方法進行研究，仍沒有辦法迅速而有效的分析其相關性。因此在本篇論文，我們提出了k-means演算法與序列分析研究理論混合型態的新式方法，來分析基因的表現值，不像以往的方法，我們的方法著重在探討調控蛋白的產生以及發生作用的時間差(time-lagged)。為了要驗證我們的效能，我們不但會在酵母菌的基因體上測試我們的方法，也會探討與細胞生長週期相關的轉錄因子以及被其調控基因之間的關係。

The expression level of genes that are co-regulated by regulatory proteins is highly correlation with the regulatory protein expressions. To understand the regulation mechanisms is crucial to system biology. However, most previous works are neither effective nor efficient. In this thesis, we propose a multi-strategy method which combines K-means with sequence alignment for gene expression analysis. Unlike most current approaches, our method address the issue of lag between regulatory proteins and target genes. To demonstrate its performance, we tested our method on the yeast genome, and also the cell cycle-related transcription factors and regulated genes.