PhD Oral Exam - Stephanie Kamgnia Wonkap, Computer Science
Gene Regulatory Network Inference Using Machine Learning Techniques
This event is free
School of Graduate Studies
When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.
Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.
System Biology is a field that aims to model complex biological systems. One of these complex systems is the gene regulatory network that plays a critical function when the organism needs to adapt to the habitat in which it lives. Understanding the factors that structure this interaction requires an elaborate analysis of the regulatory mechanisms. In a sense, a complete reconstruction of this complex gene regulatory network is required. For a long time, analysis at this level was impeded by a lack of technical development and experimental complexities. However, in recent years several technical and experimental advances have permitted the development of an unprecedented amount of data about living things. This availability of a large amount of data has stimulated a range of computational methods that model and reconstruct the gene regulatory network. The following thesis seeks to contribute to this discourse by answering the following questions: what is a gene regulatory network? Why is it important to model this complex system? What are the challenges behind gene regulatory network modeling and inference? What are the efforts that have been made so far?
Although many solutions have been proposed for the gene regulatory network (GRN) inference, it remains a central challenge in computational systems biology. The majority of existing methods do not take advantage of the complementarity of information in diverse biological data and instead concentrate on the use of unique data for the GRN inference. However, due to inherent noise and the limited availability of data commonly used by existing methods, their results have been modest. For this reason, it seems crucial to integrate diverse biological knowledge for the GRN inference. In this thesis, we propose a new algorithm BENIN that integrates different biological data to unravel the complexity of gene regulation. The algorithm is benchmarked on the DREAM4 challenge dataset and the Human Hela cell cycle gene expression dataset. On the DREAM4 dataset BENIN was perform better than the winner of one the subchallenge, and it is competitive with state of the art methods. Moreover, on Hela cell cycle data, BENIN can infer several known interactions but also news interactions that need further investigation.