Genetic regulatory networks established by shortest path algorithm and conditional probabilities for ovarian carcinoma microarray data

• Published in: 2010 International Conference on Machine Learning and Cybernetics Vol. 1; pp. 32 - 36
• Main Authors: Meng-Hsiun Tsai, Hsiao-Han Ko, Sheng-Chuan Chiu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2010
• Subjects: Bayesian methods; Bayesian theorem; Bioinformatics; Breast cancer; Dijkstra's algorithm; Gene expression; Gene regulatory network; Machine learning
• ISBN: 9781424465262; 1424465265
• ISSN: 2160-133X
• DOI: 10.1109/ICMLC.2010.5581099

In the cancer research recently, it still doesn't have a definitive conclusion for the regulatory mechanisms of tumorigenesis and metastasis. But different genes have different biological functions, and these functions with interactions between genes play an important key in gene regulatory networks. Microarray is a tool most commonly used in the disease research, and scientists usually use that the feature can accommodate huge data to record gene expressions in cancer. Then we will apply target genes to identify regulatory pathways by Dijkstra's algorithm combined with Bayesian theorem. There are 18 regulatory pathways are identified by this hybrid methods, such as chemokine signaling pathway, cell cycle and apoptosis successfully, moreover these results show the inhibition or activation between genes as well as their directions with validation by databases. This hybrid method can not only analyze complex cancer pathways, but also it will be helpful to find a more effective treatment for disease research in the future.