Negative correlation based gene markers identification in integrative gene expression data

Tao Zeng,Xuan Guo,Juan Liu

DOI: https://doi.org/10.1504/IJDMB.2014.062889

2014-01-01

International Journal of Data Mining and Bioinformatics

Abstract:long with the emergence and development of translational biomedicine, more and more genetic information has been applied in clinical practice. In the recent decade, the discovery of genetic markers for cancer prognosis obtains increasing attention and many methods have been developed. The 'element' methods use one or two independent genes to judge the Boolean status of disease. The 'set' methods use multiple gene markers as a whole to classify patients into different risks. And the advanced 'sets' methods use a group of different sets of biomarkers in an assembling manner. Either the existed 'set' or 'sets' methods only concern positive correlations among genes. However, the negative regulation, negative feedback, and functional repression are actually the relevant clues in cancer studies. Therefore, in this paper, based on the integrative gene expression data organised as gene-time-sample data or gene-sample-source data, we propose to mine Negatively Correlated Gene Sets NCGSs from multiple datasets, and use them along with the maximal positively correlated gene sets for prognosis classification. The experiment results suggest the promotions of cancer prognosis accuracy and meaningful pathogen relevance of gene markers by NCGS applications.