K. Kalantzaki, E. S. Bei, K. P. Exarchos,M. Zervakis, M. Garofalakis, and D. I. Fotiadis
This paper proposed a framework for constructing and analyzing gene-networks from sparse temporal data with the aim of understanding the mechanisms which affect the progression of oral cancer. Partial Correlations and Kernel Density Estimation network models were used in the framework to capture the genetic interactions, and a novel kernel based approach was developed to identify the genetic dependencies in the network structure. The proposed framework exploits not only the direct but also indirect genetic interactions, emphasizing on the use of the cross correlation metric in addition to the genetic causality, by means of the BIC criterion. The framework was applied to analyze tissue and blood sample data of cancer patients at different stages, and common disease-related structures were identified which may represent the association between diseases state and biological processes in oral cancer. The proposed framework of analysis provides strong evidence on the importance of an altered MET (hepatocyte growth factor receptor) network in assessing the oral cancer progression.