基於人類與小鼠微陣列基因表現圖譜識別功能性基因群

Abstract

跨物種的基因表現圖譜分析能提供在天擇的進程中被保留的基因的功能和其參與機制的訊息，保留在物種間的基因群所扮演的生化功能極可能具有不易被取代的重要功能。尋找這樣的功能性基因群能加速基因療法中候選基因的發現和藥物的開發。為此，本研究提出一個新穎的計算處理架構試圖找出保留於人類和小鼠的基因群，利用奇異值分解（singular value decomposition） 和分群演算法分析基因表現，並且利用同源連結（ortholog linkage）和時間規整演算法(time warping algorithm)使來自不同物種（異質性）的微陣列基因表現圖譜時間序列可以比較並依此建議同源基因群。同時，我們實作模糊最近聚類（fuzzy nearest-cluster）方法來預測這些可能在生物過程（bioprocess）扮演極重要的角色的同源（orthologous）基因群的功能併施行統計檢定找出具有生物意義、保留在物種間影響細胞週期的基因群。 簡而言之，這個研究結合序列和時間序列圖譜層級的相似性來建議跨物種的功能性基因群，提供基因療法實驗的候選基因並希望能貢獻在跨物種基因分析的卓越進展。 最後，為了讓整個流程方便應用在未來的相關研究上，整個架構被模組並程式化成一個獨立的可執行套件並結合視覺化的呈現。

Cross-species gene expression analysis provides information of gene functions and involving mechanisms, which conserved in evolutionary process. Gene groups conserved in species are very likely to play irreplaceable biochemical functions. Searching for this kind of functional gene groups can accelerate the discovery of candidate genes in gene therapy and development of drug design. For this, our research proposes a novel computational scheme to figure out the genes having important biochemical functions, especially targets on the genes which are conserved in human and mouse. These genes conserved across evolutionary history would be most likely to reveal fundamental biochemical functions. This work utilizes singular value decomposition (SVD) and clustering techniques to analyze gene expression, and exploits orthologous linkage and time warping algorithm making microarray time-series gene-expression profiles of different species (heterogeneous profiles) comparable to suggest orthologous gene groups. In the meanwhile, in order to make the results more promising, we use fuzzy nearest cluster method to predict the functions of orthologous genes which might play important roles in the bioprocess iii and perform statistical test according to our annotation of predicated gene function to find the genes having biological significance among these orthologous genes. In brief, this research combines sequence- and time-series expression- levels ortholog to suggest functional genes among multiple species, provides materials for candidate gene therapy experiments and hopes to contribute remarkable advancement in cross-species orthologous gene analysis. In the end, in order to let the whole process be utilized in further application, the scheme is modeled and programmed in a standalone executable package with visualized presentation.