由基因本體論分群探勘基因調節網路

Abstract

基因調節網路可以幫助生物學家了解更多系統性的生物現象。在本論文中，我們整合時間序列生物晶片的資訊和基因本體論，並提出了一個資料探勘的方法以尋找基因群之間的調節網路。 首先，我們將時間序列生物晶片的資料轉換成基因改變傾向的資料，並利用基因本體論對基因的註解，將基因分成數個基因群。對於每個基因群，我們尋找基因群內的調節樣式。然後由這些調節樣式，我們尋找基因群和基因群之間包含與相反的調節樣式，以推論基因群之間的調節關係，並建立基因群之間的調節網路。 實驗結果顯示我們所提出的方法具有效率性與擴充性，可以讓我們從一個全觀的角度去了解基因調節網路。我們提出的方法不只可以找到一些被生物學家驗證過的調節關係，同時也可以找到一些新的調節關係須要由生物學家進行進一步的確認。

The gene regulation networks can help biologists know more about the systematical biological phenomena. In this thesis, we integrate time-series microarray and the Gene Ontology, and propose a data mining approach to find gene regulation networks between gene categories. We first transform the time-series microarray dataset into gene tendency profiles and use the Gene Ontology annotations to classify genes into gene categories. For each gene category, we find its regulation patterns. By using the regulation patterns found for each gene category, we can infer the gene regulation relationships by finding the inclusive and opposite patterns between gene categories. Base on the regulation relationships inferred, we can construct gene regulation networks. The experiment results show that our proposed method is efficient and scalable. Our method can provide a global view of gene regulation networks, which include not only some meaningful regulation relationships verified by biologists, but also some regulation relationships share regulation patterns, which need to be further verified by biologists.