Identification of plausible candidates in Prostate Cancer using integrated machine learning approaches

Abstract

Abstract Background Prostate-specific antigen (PSA) in present times is a widely used Prostate Cancer (PCa) biomarker. PSA is associated with some variables that often turn out to be a false positive result or even end up in unnecessary biopsies of older people. Methods Extensive literature survey was done, and some clinical parameters were taken for its associated comorbidities like diabetes, obesity, and PCa. These parameters were selected considering how the deviation in their threshold values could accelerate the complex process of carcinogenesis, more specifically corresponding to PCa. The collected data was converted to semi-binary data (-1, -0.5, 0, 0.5, and 1), wherein we aim to apply machine learning (ML) algorithms. For the identification of candidate biomarkers, first, we cross-checked different publicly available datasets some published RNA-seq datasets and our own whole-exome sequencing data to identify common role players among PCa, diabetes, and obesity. Interactome networks were analyzed using GeneMANIA and visualized using Cytoscape to narrow down their common interacting partners, and later cBioportal was used (for comparing expression level analysis based on Z scored values) wherein different types of mutation w.r.t their expression and mRNA expression (RNA seq FPKM) plots. GEPIA 2 tool was applied to see the difference in expression of resulting commonalities between the normal tissue and TCGA datasets of PCa. Top ranking genes were selected to show vivid clustering coefficients with help of the Cytoscape-cytoHubba plugin and for ascertaining survival plots GEPIA 2 is used. Results Comparing different publicly available datasets, we get BLM as a common player among all the three diseases, whereas when publicly available datasets, GWAS dataset, and published sequencing results were compared, SPFTPC and PPIMB were the most common. TMPO and FOXP1 were identified as common interacting partners with the help of GeneMANIA and are also seen interacting with BLM. Conclusions A probabilistic machine learning model was achieved to identify key candidates between Diabetes, Obesity, and PCa. This, we believe would herald precision scale modeling for easy prognosis