Functional Analysis and Characterization of Differential Coexpression Networks

Differential coexpression analysis is emerging as a complement to conventional differential gene expression analysis. The identified differential coexpression links can be assembled into a differential coexpression network (DCEN) in response to environmental stresses or genetic changes. Differential coexpression analyses have been successfully used to identify condition-specific modules; however, the structural properties and biological significance of general DCENs have not been well investigated. Here, we analyzed two independent Saccharomyces cerevisiae DCENs constructed from large-scale time-course gene expression profiles in response to different situations. Topological analyses show that DCENs are tree-like networks possessing scale-free characteristics, but not small-world. Functional analyses indicate that differentially coexpressed gene pairs in DCEN tend to link different biological processes, achieving complementary or synergistic effects. Furthermore, the gene pairs lacking common transcription factors are sensitive to perturbation and hence lead to differential coexpression. Based on these observations, we integrated transcriptional regulatory information into DCEN and identified transcription factors that might cause differential coexpression by gain or loss of activation in response to different situations. Collectively, our results not only uncover the unique structural characteristics of DCEN but also provide new insights into interpretation of DCEN to reveal its biological significance and infer the underlying gene regulatory dynamics.


Figure S2. Illustration of triads in coexpression networks
There are four possible types of triads according to the combinatorial patterns of the three interconnected signed links (A). However, only type 1 and 2 triads can be observed in coexpression network because of the transmission characteristics of correlation. Because coexpression networks contain this specific property, no complete differential coexpression triad can be obtained (B). The red and green links denote the positive and negative correlation pairs. The blue link represents the significantly differential coexpressed link.

Figure S3. Distributions of transitivity in differential coexpression networks
The transitivity of a graph T is defined as , where λ is the number of triads and τ is the number of triples. A triad is a complete subgraph with exact three nodes, and a triple is a subgraph with three nodes and two edge. The blue histogram represents the random distribution of transitivity of 100,000 randomized differential coexpression networks (DCENs) and Red line denotes the observed value of the given DCEN. A randomized DCEN was derived from gene expression data which relations between genes and expression profiles in a condition are randomized.

Figure S4. Differential coexpression analysis of strong coexpression triads.
A strong coexpression triad is defined as a triad in which expression profiles among genes are highly correlated in one condition (Spearman correlation coefficient > 0.9), and then we examined if these strong triads lose coexpression in another condition (A). The percentages of strong triads with loss of coexpression derived from dataset 1 and 2 are illustrated in (B) and (C), respectively. Red and blue lines represent the results from real and randomized gene expression data, respectively.

Figure S5. Topological properties of differential coexpression networks constructed by DCe
The method DCe, implemented in R package DCGL, was applied to the two gene expression datasets, respectively. The used parameters are as follows: link filtering method is "qth"; cutoff used for link filtering is 0.1; correlation coefficient is computed by spearman method; and the cutoff of q-value is 0.01.

Figure S6. Cumulative distributions of expression correlation of TF co-targets under different glucose concentration
For each TF (Table2), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the p-value of Kolmogorov-Smirnov test.

Figure S6. (cont.) Cumulative distributions of expression correlation of TF co-targets under different glucose concentration
For each TF (Table2), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the p-value of Kolmogorov-Smirnov test.

Figure S6. (cont.) Cumulative distributions of expression correlation of TF co-targets under different glucose concentration
For each TF (Table2), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the p-value of Kolmogorov-Smirnov test.

Figure S6. (cont.) Cumulative distributions of expression correlation of TF co-targets under different glucose concentration
For each TF (Table2), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the p-value of Kolmogorov-Smirnov test.

Figure S7. Cumulative distributions of expression correlation of TF co-targets in wide type and YOX1/YHP1 mutant
For each TF (Table 3), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the P-value of Kolmogorov-Smirnov test.

Figure S7. (cont.) Cumulative distributions of expression correlation of TF co-targets in wide type and YOX1/YHP1 mutant
For each TF (Table 3), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the P-value of Kolmogorov-Smirnov test.  (Table 3), the expression correlations of pair-wise combinations of the genes which are target by the give TF are presented. The D is the distance between two cumulative distributions and the P is the P-value of Kolmogorov-Smirnov test.