Application of A Novel Triclustering Method in Analyzing Three Dimensional Transcriptomics Data

by Anirban Bhar
Doctoral thesis
Date of Examination:2015-03-24
Date of issue:2015-06-19
Advisor:Prof. Dr. Edgar Wingender
Referee:Prof. Dr. Edgar Wingender
Referee:Prof. Dr. Stephan Waack
Referee:Prof. Dr. Burkhard Morgenstern
Referee:Prof. Dr. Anita Schöbel
Referee:Prof. Dr. Tim Beißbarth
Referee:Prof. Dr. Dieter Hogrefe
Persistent Address: http://dx.doi.org/10.53846/goediss-5150

 

 

Files in this item

Name:Anirban_PhD_thesis_final_1.pdf
Size:14.3Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

Due to the advancement of microarray technology over the last decade, it is feasible to monitor the gene expression dynamics not only over a set of replicates but also either a set of time points or doses of chemical substances. In such three dimensional datasets, variations in the expression profiles can not only be observed across the time points or doses of the chemical substances but also across the replicates due to either abnormalities in the experimental protocol or the physiological variations. Thus, it is important to mine such three dimensional datasets in order to extract biologically meaningful information. In this work, I have proposed a novel triclustering algorithm δ-TRIMAX by introducing a mean squared residue (MSR) score as a coherence measure of the resultant triclusters. The application of this algorithm has been shown in the context of breast cancer progression in order to reveal potential biological processes driving breast cancer invasion. Moreover, I have proposed an improved version of δ-TRIMAX, the EMOA-δ-TRIMAX algorithm which effectively deals with the pitfalls of the former one. One artificial dataset and three real-life datasets have been used to compare the performance of the proposed algorithms with that of other existing algorithms. Besides, the improved version has been applied to one dataset monitoring expression profiles of genes during breast cancer progression for unveiling regulatory mechanisms. Furthermore, the application of the EMOA-δ-TRIMAX algorithm has been demonstrated in investigating the potential biological processes and transcriptional regulatory mechanisms involved in the adolescence of cardiomyocytes. Additionally, I have applied EMOA-δ-TRIMAX algorithm to four real-life datasets in order to provide hints on the pathways perturbed by different toxicants in different tissues. Overall, I could demonstrate that the results of the proposed algorithms for each of the real-life datasets and the artificial ones are promising and provide new insights into the context of breast cancer progression, cardiomyocytes generation and explaining inhalation toxicity.
Keywords: Bioinformatics; Co-expression; Co-regulation; Developmental Biology; Gene Regulatory Network
 
Statistik