Abstract:
Recent advances have enlightened that biological pathways are far more complicated than once thought, due to the inclusion of interconnected complex cellular actions, which made hard understanding the multifaceted mechanisms behind the biological phenomena. As a panacea, the bioinformatics community has brought up the modularity concept to ease the understanding of biological ground truth. A microarray is a high-throughput technology, which provides a global view of the genome in a single experiment with a systematic manner by enabling the analysis of the expression levels of a large number of genes simultaneously. Bayesian networks are probabilistic graphical models, which are well proven technique to infer gene regulatory networks from microarray data because of their ability to incorporate prior knowledge. In this study, we present an algorithm, called BNP, to infer biological pathways. Fortifying the results obtained by our model and exploring the novel interactions between genes, we construct a gene interaction atlas via Bayesian networks by incorporating external biological knowledge. Furthermore, a comparison of our methodology with the FLAT method, which does not use any external knowledge, shows that BNP outperforms it in all simulations.
