Inferring transcriptional regulations from temporal properties of gene expression

Abstract

In an effort to identify localized causal signal between the gene expression profiles of a regulator and its targets, a new method of inferring transcriptional regulations between genes has been proposed in this thesis. The proposed method utilizes time series gene expression data and identifies the localized causal signal using local time warped alignment. The method employees time warping technique to mitigate the error that often results from inconstant time delay of the target gene``s response. Performance of the method has been tested and compared with two other conventional methods (i.e., Pearson correlation and event method) by analyzing known transcriptional regulation data set. The result of the test revealed that; i) both the proposed method and Pearson correlation identified significant number of true pair in high ranked ones; ii) the proposed method remarkably well detected transcriptional regulation relationships having inconstant time delay; and iii) detailed information related to the temporal transcriptional regulation such as the timing and duration beyond the score can be extracted from the localized causal signal obtained from the proposed method.