Table 1.

Methods for constructing gene co-expression networks.

Algorithms and applications	Advantages	Potential limitations
Quasi-Clique Merger algorithm for finding co-expressed gene clusters48.	Integrates multiple microarray datasets, even including the data without normal samples48.	Requires large datasets to ensure a high level of significance for correlations of gene expressions.
Context Likelihood of Relatedness algorithm for inferring edges in the network to identify cross-species gene interactions50.	Captures nonlinear changes in gene expressions50.	Can’t discriminate the direction of correlations of gene pairs without Pearson Correlation Coefficient.
GENIE3 for inferring gene co-expression network49.	Fast detects gene networks from large multifactorial gene expression data.	Requires prior knowledge of the transcription factors.
WGCNA for detecting functional gene clusters47.	The approximately scale-free network structure reserves connectivity when randomly removing nodes.	Sensitive to the number of genes and the choices of parameters (i.e., soft threshold).