A method for gene essentiality in miRNA-TF-mRNA co-regulatory network and its application on prostate cancer

Prostate cancer is dangerous due to its metastatic property in the advanced stage. Considerable advance has been made in unraveling the tumor-related molecular basis. Traditionally, differentially expressed genes are focused on for biomarker identification. Here, we argued that subtle gene alterations could be amplified through the biological network, those omissive non-significantly disrupted genes might be also cancer-related. To verify our hypothesis, we constructed a co-regulatory network comprised of miRNA, TF and mRNA. Then an essential score for a gene was computed by summing its neighbor expression weighted by the Jaccard similarity coefficient. The genes with the most essential score alterations (ESA) were selected for the following analyses. Although most of these genes exhibit only subtle alterations at expression levels, they are indeed informative in discriminating between tumor and normal samples. Moreover, the enrichment analysis revealed they are cancer-associated. We, therefore, debated that they could be used as cancer-related factors. Finally, feed-forward loops (FFLs) were constructed based on these genes for further investigating their potential regulation patterns associated with the prostate cancer. We found most nodes in the FFLs related to cancer. Our results suggested the subtle alterations could impact the biological system by their interaction with other elements. This observation would shed light on deep understanding the mechanism of the prostate cancer.