Abstract

Abstract: Discovering the human genes that cause disease (or ``disease genes”) is one of the emerging tasks in biomedicine informatics. In many ongoing research projects, protein-protein interaction networks (PPI networks) are being exploited in the discovery process, because the network-neighbor of a disease gene is likely to cause the same or the similar disease. After an introduction to the complex relationship between disease genes and PPI networks, we will give a brief overview of various computational methods in discovering disease genes based on PPI networks. Finally, we will present an effective method to predict disease genes, by taking advantage of the wealth of existing data of disease gene neighbors which often contain information about as-yet-unidentified disease genes. Our method uses semi-supervised learning, integrating data of human PPI networks and various biological data extracted from multiple proteomic and genomic data sources. The experimental results demonstrated that our proposed method performed well with high accuracy, and at the same time, predicted some putative disease genes. Performing the experimental results with small amounts of labeled data, the output demonstrated an improved ability to study specific diseases when the known disease genes (as labeled data) are very often limited.

Biography: Phuong Nguyen received her B.S. (2003) and M.S. (2005) in Information Technology from Hanoi University of Technology (HUT), Vietnam. From 2003 until the present, she was a lecturer at the Faculty of Information Technology, HUT . In September 2008, she received a PhD in Bioinformatics from the School of Knowledge Science at the Japan Institute of Science and Technology (JAIST). Her current research interests are formal model, data mining and machine learning applied to biological and medical data in general, and molecular networks and molecular networks related to diseases in particular. Phuong joined CoSBi in October 2008.