利用階層式叢集及不同分類方法分析人類正常組織特異性基因

Using hierarchical clustering and different classification methods to analyze human normal tissue specific genes

陳尚宏(Shang-hung Chen)

指導教授 ： 王孫崇

國立中央大學/理學院/系統生物與生物資訊研究所/碩士(2011年)

摘要

隨著生物資訊領域越來越蓬勃發展，微陣列晶片也被廣為應用，我們的研究即是使用微陣列晶片資料來做後續分析，資料的來源是根據BioGPS這個資料庫所提供經過正規化處理後的晶片資料，這些晶片資料也已上傳到NCBI所屬的GEO，GEO這個資料庫有著各種不同的晶片實驗資料，而我們所著重的目標在於找出人類正常組織具有特異性的基因，找出上述的基因之後我們可以知道每個組織或細胞型態的主要表現基因是哪些，接著我們想更深入了解這些人類正常組織間彼此之間的相對應關係，所以我們將這些特異性基因依照不同情況再利用階層式集群分析法來做分群，並且使用了數種不同的分類方式嘗試去解釋這些分群的結果。
從我們的研究中利用兩種不同的篩選組織特異性基因的方法，第一個篩選方法是找出在不同組織及細胞型態間表現量具有較大變異的基因，第二個篩選方法是參考現存論文的篩選方式，找出每個組織及細胞型態各自特定高表現的基因，條件限制較多，還需符合基因在不同組織及細胞型態間的最高表現量與次高表現量的比值大於兩倍，從我們的結果發現利用第一種篩選組織特異性基因的方法在不論是發育學的胚層概念、生理學、五臟六腑以及十二經絡等四種不同的分類方式所做的階層式叢集分群都會是比較一致的；確定方法之後，我們進一步探討發現組織特異性基因的表現量所做的階層式叢集分群也會比用組織特異性基因在染色體上的分布情形來的理想許多；最後中醫與西醫之間的比較，中醫與西醫的觀點對於解釋這些階層式叢集的分群結果其實是難分軒輊、不分上下的。

關鍵字

數種不同的分類方式 ； 階層式叢集分析法 ； 人類正常組織特異性基因 ； 微陣列晶片

並列摘要

With more and more vigorous development of the field of bioinformatics, microarray chips have been widely used, our study is the use of microarray chips to do the following analysis, the data source is based on BioGPS, and it provides normalized processing chip data, these chips data have also been uploaded to the GEO belongs to NCBI, this database has a variety of chip experimental data, our goal is to identify human normal tissue specific genes, when we identify these genes, we can know each tissue or cell type which genes are mainly express, then we want to deeply understand the corresponding relationship between these human normal tissues, so we take these specific genes based on different conditions and next use hierarchical clustering analysis method to do cluster, and then we use several different methods of classification and try to explain these grouping results.
From our study, we using two different screening methods of tissue-specific genes; the first screening method to identify genes with large expression levels variation between different tissues and cell types, the second screening method we take a reference to the existing paper, identify each tissue and cell type their specific high expression genes, ,with more conditions restriction, it also has to meet the genes between different tissues and cell types highest expression level division the second high expression level ratio of greater than two times, from the results, we find the first screening method of tissue-specific genes hierarchical clustering would be more consistent in the four classification of developmental germinal layer concept,physiology, five zangs and six fus, and the twelve main meridians; after determination of the method, we further explore the use of tissue specific genes expression levels hierarchical clustering results are also more ideal than use of tissue specific genes on the chromosome’s distributions; at last, we compare Chinese medicine and Western medicine, Chinese medicine and Western medicine points of view to explain the hierarchical clustering results is nck and neck.

並列關鍵字

several different methods of classification ； hierarchical clustering ； human normal tissue specific genes ； microarray chip

參考文獻

2. Wu, C., et al., BioGPS: an extensible and customizable portal for querying and organizing gene annotation resources. Genome Biol, 2009. 10(11): p. R130.

3. Barrett, T., et al., NCBI GEO: mining tens of millions of expression profiles--database and tools update. Nucleic Acids Res, 2007. 35(Database issue): p. D760-5.

4. Ge, X., et al., Interpreting expression profiles of cancers by genome-wide survey of breadth of expression in normal tissues. Genomics, 2005. 86(2): p. 127-141.

5. Ishwaran, H. and J. Sunil Rao, Clustering gene expression profile data by selective shrinkage. Statistics & Probability Letters, 2008. 78(12): p. 1490-1497.

6. Xiao, S.J., et al., TiSGeD: a database for tissue-specific genes. Bioinformatics, 2010. 26(9): p. 1273-5.