CoXpress: differential co-expression in gene expression data

Abstract

Background: Traditional methods of analysing gene expression data often include a statistical test to find differentially expressed genes, or use of a clustering algorithm to find groups of genes that behave similarly across a dataset. However, these methods may miss groups of genes which form differential co-expression patterns under different subsets of experimental conditions. Here we describe coXpress, an R package that allows researchers to identify groups of genes that are differentially co-expressed.

Results: We have developed coXpress as a means of identifying groups of genes that are differentially co-expressed. The utility of coXpress is demonstrated using two publicly available microarray datasets. Our software identifies several groups of genes that are highly correlated under one set of biologically related experiments, but which show little or no correlation in a second set of experiments. The software uses a re-sampling method to calculate a p-value for each group, and provides several methods for the visualisation of differentially co-expressed genes.

Conclusion: coXpress can be used to find groups of genes that display differential co-expression patterns in microarray datasets.

Figure 1

Conceptual flow of analysis through the coXpress function. This figure shows the conceptual steps taken by the coXpress function in order to find differentially co-expressed groups of genes in microarray data. Steps should be followed in numerical order.

Figure 2

A differentially co-expressed group from the Golub dataset compared to random distributions. Group 3 (n = 34) from the Golub [21] data compared to random distributions. A) the distribution of pairwise correlation coefficients for group 3 in the ALL subset (blue) compared to the distribution of pairwise correlation coefficients from a group of the same size generated by the random uniform distribution (red) and the distribution of pairwise correlation coefficients from a group of the same size randomly selected from the dataset (green). B) Equivalent graph to A for the AML subset. C) A comparison of the observed t-statistic for group 3 in the ALL subset with a distribution of 10,000 random t-statistics generated by randomly resampling the dataset. D) Equivalent graph to C for the AML subset. Distributions were smoothed and drawn using density function in R [16]. Note that graph C has a broken x-axis.

Figure 3

Expression profiles for three differentially co-expressed groups in the Golub dataset. Expression profiles for three groups of differentially co-expressed genes from the Golub dataset [21]. A) Group 3 (n = 34) in 27 ALL samples (left) and 11 AML samples (right). B) Group 62 (n = 7) in 27 ALL samples (left) and 11 AML samples (right). C) Group 121 (n = 12) in 27 ALL samples (left) and 11 AML samples (right). Expression levels have been scaled and centred.

Figure 4

Images of the correlation matrices for three differentially co-expressed groups in the Golub dataset. Images of the correlation matrices for three groups of differentially co-expressed genes from the Golub dataset [21]. A) Group 3 (n = 34) in 27 ALL samples (left) and 11 AML samples (right). B) Group 62 (n = 7) in 27 ALL samples (left) and 11 AML samples (right). C) Group 121 (n = 12) in 27 ALL samples (left) and 11 AML samples (right). Each coefficient in the correlation matrix is represented as a square, with the colour of the square representing the amount of correlation. The colour scale used is green to red, with green representing -1, red representing +1 and black representing 0.

Figure 5

Expression profiles for three differentially co-expressed groups in the ALL subtype dataset. Expression profiles for three groups of differentially co-expressed genes from the Yeoh et al dataset [23]. A) Group 47 (n = 16) in 15 BCR-ABL samples (left) and 43 T-ALL1 samples (right). B) Group 31 (n = 10) in 15 BCR-ABL samples (left) and 43 T-ALL1 samples (right). C) Group 89 (n = 13) in 15 BCR-ABL samples (left) and 43 T-ALL1 samples (right). Expression levels have been scaled and centred.