Gene set-based integrative analysis of ovarian clear cell carcinoma

The pathogenesis of ovarian clear cell carcinoma is still poorly understood; therefore, we conducted a gene set-based analysis by integrating datasets downloaded from publicly available microarray gene expression databases to investigate the pathogenesis of clear cell carcinoma, which was based on the regularity of functions defined by gene ontology or canonical pathway databases.

The gene expression profiles of 80 clear cell carcinomas and 136 normal ovarian controls were downloaded from the National Center for Biotechnology Information Gene Expression Omnibus database. The gene expression profiles were converted to the gene set regularity (GSR) indexes computed using the modified differential rank conservation, an algorithm measuring the degree of gene expression ranking change in a gene set. Then the differences of GSR indexes between clear cell carcinomas and normal ovarian controls were analyzed.

Machine learning can accurately recognize and classify the patterns of functional regularities containing the GSR indexes between the clear cell carcinomas and normal controls with an accuracy of 99.3%. The significant aberrations included oxidoreductase activity, binding, transport, channel activity, cell adhesion, immune response, chromosome assembly, and the deregulated signaling molecules, such as guanyl nucleotide exchange factors, phosphoinositide 3-kinase-activating kinase, receptor tyrosine kinase B, and protein tyrosine kinase.

Our pioneering works using the functionome, which was converted from microarray gene expression profiles for integrative analysis, showed a clear distinction of functional changes between the clear cell carcinomas and normal ovarian controls. This approach might provide a comprehensive view of the deregulated functions of clear cell carcinomas for further investigation.