生物晶片探針設計□ 利用16S核糖體RNA作細菌分類

Abstract

應用生物晶片來檢測特定微生物菌種，相較於傳統微生物檢測法具有快速、簡便及鑑別力更佳等優勢，因此在醫學診療、農業、國防與環境汙染控制方面極具裨益。在本論文中，我們提出應用決策樹演算法及基因規畫，對微生物的16S rRNA序列，設計出可供鑑別特定菌種的探針集合，意即為某一細菌或某特定細菌族群找到特殊片段做為識別標誌(diagnostic signature)。除了對已知微生物菌種作檢測外，我們希望藉已知的微生物序列集合來預測未知微生物菌種在微生物種系發展世系圖(phylogenetic tree)上的位置，而能更進一步預測此微生物在系統分類學上的「種」、「屬」層次。然而，由於微生物的16S rRNA序列具有高度相似性(homology)，要為無法用單一生物片段作識別標誌的微生物菌種，找出識別標誌，我們提出更具有表達能力的表示法來設計探針。由於生物晶片的探針個數會影響其成本，我們將此問題視為一最佳化的議題，不僅要能找出可供鑑別、預測的探針集合，更期望能找到探針個數最少的識別標誌。此外，為了提高生物晶片的準確性，本論文同時考慮到生物序列雜交反應(hyberdization)的特性，才能設計出具有專一性及敏感性的良好探針。

Compared with traditional microorganism identification technology, using DNA chip to identify specific microorganism is much faster, convenient and powerful. Therefore, this technology is a great benefit in the fields of medical science, agriculture, national defense, and environmental monitoring. In this thesis, we propose that by decision tree algorithm and genetic programming, we can design probe set to identify specific bacterium according to their 16S rRNA sequences, i.e. search out significant characteristic DNA segment to be diagnostic signature for each bacterium or specific bacterium family. We intend to identify known bacterium; moreover, hope that we can predict the position of phylogenetic tree for unknown microorganism by the known microorganism sequence data set, and then we can further predict which genus or species this microorganism is. However, since microorganism’s 16S rRNA sequences are highly homologous, to search diagnostic signature for each microorganism, we need a more powerful expressiveness way to interpret. We take this problem as a optimal issue because the number of DNA chip probes will influence the manufacture cost. Therefore, we need design probe set not only that have distinguishability and prediction ability, but also which number is as low as possible. Besides, to enhance the quality of DNA chip, the property of hybridization must be considered, and then the good probes that are specific and sensitive can be designed.