A novel density based community detection algorithm and its application in detecting potential biomarkers of ESCC

Abstract

The development of statistically and biologically competent Community Detection Algorithm (CDA) is essential for extracting hidden information from massive biological datasets. This study introduces a novel community index as well as a CDA based on the newly introduced community index. To validate the effectiveness and robustness of the communities identified by the proposed CDA, we compare with six sets of communities identified by well-known CDAs, namely, FastGreedy, infomap, labelProp, leadingEigen, louvain, and walktrap. It is observed that the proposed algorithm outperforms its competing algorithms in terms of several prominent statistical and biological measures. We implement the hardware coding with Verilog, which surprisingly reduces the computation time by 20% compared to R programming while extracting the communities. Next, the communities identified by the proposed algorithm are used for topological and biological analysis with reference to the elite genes, obtained from Genecards, to identify potential biomarkers of Esophageal Squamous Cell Carcinoma (ESCC). Finally, we discover that the genes F2RL3, CALM1, LPAR1, ARPC2, and CLDN7 carry significantly high topological and biological relevance of previously established ESCC elite genes. Further the established wet lab results also substantiate our claims. Hence, we affirm the aforesaid genes, as ESCC potential biomarkers.