Integration of Automatic Text Mining and Genomic and Proteomic Analysis to Unravel Prostate Cancer Biomarkers.

Abstract

Prostate cancer (PCa) is the most prevalent noncutaneous cancer among men. The limited accuracy and/or invasive nature of the current diagnostic tools have driven the demand for new and noninvasive biomarkers. Urine as a noninvasive sample that contains prostatic secretions is a promising source of PCa markers. The automatic text-mining functionality of VOSviewer was used to retrieve and create co-occurrence networks of terms associated with PCa. These results were complemented with DisGENET data, a repository of PCa associations, and with a recent bioinformatic analysis integrating all differentially expressed proteins identified in tumor tissue and urine from PCa patients to address the limited term selection of VOSviewer. Afterward, the results were integrated with gene expression data from the Gene Expression Omnibus database to correlate gene and protein levels. This study suggests AXIN2, GSTM2, KLK3, LGALS3, MSMB, PRTFDC1, and SH3RF1 as important entities in PCa context. KLK, LGALS3, and MSMB proteins are common to a previous bioinformatic analysis, and a concordance was found between the levels of gene and protein expression. The applicability of the pipeline presented here was validated by showing altered urinary levels of galectin-3 protein in PCa patients compared to noncancer subjects.