A Weighted Gene Co-Expression Network Analysis Based on Iterative Learning

Abstract

Gene co-expression networks are widely used in field of exploring gene function and describe connections not only between genes in the network, but also between gene and trait of biological significance. As a part of building gene co-expression network, the adjacency function plays an important role in describing the connections between genes, which converts the co-expressive measure of each gene pair to its connective weight. A suitable parameter of adjacency function is determined by a biologically motivated criterion known as the scale-free topology fitting index in the gene co-expression network. In generally, the parameter of adjacency function is selected by manually looking up parameter table that is generated with the R software package WGCNA (Weighted Gene Co-Expression Network Analysis - WGCNA). Therefore, the revised algorithm of selecting parameter of adjacency function is proposed. Iterative learning algorithm is used to automatically find the required parameter of adjacency function and the Mouse module-trait relationships are constructed.