利用轉錄因子活性機率樣式預測轉錄調控關係

A New Approach to Identifying Transcriptional Regulatory Relationship Based on Transcription Factor Activity Probability Pattern

蔡宗曄(Tsung-Yeh Tsai)

指導教授 ： 陳中明

國立臺灣大學/工學院/醫學工程學研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.01556

摘要

基因與生命現象間的關係在高通量分生技術高度發展的今日，得以有系統且全面性地進行探討與研究。對於鉅量而高雜訊的高通量訊息，生物資訊研究提供有效率的分析方法，促使分子生醫研究在後基因體時代之蓬勃發展。本研究以生物資訊之角度出發，針對基因體學中轉錄調控關係探討之議題，整合高通量訊息與生醫文獻，提出嶄新之模擬概念與關係預測演算法。
生命體以基因表現產物為各種行為與反應之作用者，而基因轉錄表現量的變化主要受到轉錄因子促進與抑制作用之調控，因此轉錄調控關係資訊可視為轉錄因子與生命現象之連結，能協助生醫研究在生化途徑、基因醫療等方面之發展。
染色體免疫沉澱晶片是目前少數能提供全面性轉錄調控關係訊息的方法，但其有高雜訊與顯著程度定義之問題，因此許多整合染色體免疫沉澱晶片與其他多元資料的關係預測生物資訊演算法被提出。然而其中多數研究對轉錄因子與受控基因兩者表現量間的關係，皆有悖於生物真實現象之假設，本研究即針對此問題，提出符合真實調控現象之關係預測演算法。
本研究先以小波除噪與二元化方法取得基因表現量顯著變化之時間序列，再提出轉錄因子活性機率樣式的概念，根據染色體免疫沉澱晶片資料篩選學習資料，推估轉錄因子在各時間點具調控活性之機率，並根據兩時間序列之相關性，以統計檢定方式判斷其調控型態與顯著性。
將之應用於酵母菌細胞週期資料，轉錄因子活性機率樣式符合已知作用階段訊息，且以十折交叉驗證顯示具有高度強健性；關係預測演算法則在十折交叉驗證和四筆高可信度驗證資料測試下，證明具有良好的預測能力，而對非細胞週期作用轉錄因子亦能維持其預測能力。最後並應用於酵母菌熱休克實驗資料，同樣具有強健性與高準確率，顯示本方法應用之廣度。

關鍵字

轉錄因子 ； 調控關係 ； 轉錄因子活性機率樣式 ； 因素分析 ； 小波去噪

並列摘要

To depict real phenomenon of transcriptional regulation and improve the accuracy of identifying regulatory relation, a new serial pattern, called Transcription Factor Activity Probability Pattern (TFAPP) and prediction method are developed in this study. Transcriptional regulation is the keystone of biological systems; therefore, the regulatory relationship information is helpful to researches on biological pathway and genetic therapy.
We present TFAPP, the occurrence probability of regulation based on the co-expression phenomenon of targeted genes regulated by one TF, and identify regulatory relationship according to the correlation between TFAPP and binary gene expression pattern. Learning data of TFAPP are collected from ChIP-chip and microarray, and pre-processed by wavelet de-noise. TFAPP is estimated by binary factor analysis and random sampling process.
In this research, the TFAPP has been proved meaningful for 37 yeast TFs in cell cycle condition, according to most well-known information, and its robustness has also been confirmed by 10-fold cross validation. High accuracy of prediction method based on TFAPP is validated by four high-confident targeted genes lists. In summary, the successful validation and application results of TFAPP with multiple conditions data reveal the extensity and potentiality of this study.

並列關鍵字

Transcription factor ； Transcriptional regulation ； Transcription Factor Activity Probability Pattern ； Factor analysis ； Wavelet de-noise

參考文獻

Balasubramaniyan, R., E. Hullermeier, et al. (2005). "Clustering of gene expression data using a local shape-based similarity measure." Bioinformatics 21(7): 1069-77.

Beal, M. J., F. Falciani, et al. (2005). "A Bayesian approach to reconstructing genetic regulatory networks with hidden factors." Bioinformatics 21(3): 349-356.

Buck, M. J. and J. D. Lieb (2004). "ChIP-chip: considerations for the design, analysis, and application of genome-wide chromatin immunoprecipitation experiments." Genomics 83(3): 349-360.

Chen, K. C., T. Y. Wang, et al. (2005). "A stochastic differential equation model for quantifying transcriptional regulatory network in Saccharomyces cerevisiae." Bioinformatics 21(12): 2883-90.

Chuang, C. L., C. H. Jen, et al. (2007). "A Pattern Recognition Approach to Infer time-lagged Genetic Interactions." Bioinformatics(submitted).