Effective utilisation of influence maximization technique for the identification of significant nodes in breast cancer gene networks

Abstract

Background: Identifying the most important genes in a cancer gene network is a crucial step in understanding the disease's functional characteristics and finding an effective drug.

Method: In this study, a popular influence maximization technique was applied on a large breast cancer gene network to identify the most influential genes computationally. The novel approach involved incorporating gene expression data and protein to protein interaction network to create a customized pruned and weighted gene network. This was then readily provided to the influence maximization procedure. The weighted gene network was also processed through a widely accepted framework that identified essential proteins to benchmark the proposed method.

Results: The proposed method's results had matched with the majority of the output from the benchmarked framework. The key takeaway from the experiment was that the influential genes identified by the proposed method, which did not match favorably with the widely accepted framework, were found to be very important by previous in-vivo studies on breast cancer.

Interpretation & conclusion: The new findings generated from the proposed method give us a favorable reason to infer that influence maximization added a more diversified approach to define and identify important genes and could be incorporated with other popular computational techniques for more relevant results.

Keywords: Cancer gene networks; Gene expression dataset; Influence maximization; SIMPATH algorithm; WGCNA package; Weighted PPI networks.