探索擁有多調控序列的調控模組

Abstract

調控機制為生命運作中最為複雜的秘密。基本上，調控機制是由轉錄因子與基因的上游區域做結合，進而控制基因表現而引響生物體的生命運作。我們再此提出另一種模擬的模型來呈現多個轉錄因子之間的交互作用，與之前的研究不同的是，我們不但考慮結合區域的一致性，也加入了機率模型來描述不同轉錄因子交互作用時其之間的間距。利用貝氏架構的推論，我們將結合區域的一致性以及其之間間距的兩大考量做結合。我們將此演算法，SAMLA，測試大腸桿菌上的資料，並與其他的演算法作比較，可以發現 SAMLA 的確能更加精準的預測大腸桿菌中的調控模組。

One of the keys to deciphering the secrets of life is to understand transcriptional regulation mechanisms. Such mechanisms are typically mediated by the binding of transcription factors to specific upstream or downstream regions of genes, which leads to most recent studies focused on the conservation of binding sites. Unlike current research, we address the importance of the distance between binding sites for the prediction of Regulatory Modules. Based on the Bayesian framework, we present an algorithm, SAMLA (Simulated Annealing for Multiple Local Sequences Alignment), which takes into account the consensus levels of biding sites as well as the relative distance among them. To demonstrate the performance of our new approach, we conducted a comparative study with several current methods. We tested them on the datasets derived from E.coli, and the experimental results show that our method significantly outperforms the others.