Skip to main content

Advertisement

Figure 5 | BMC Bioinformatics

Figure 5

From: GSVA: gene set variation analysis for microarray and RNA-Seq data

Figure 5

Comparison of differential pathway activity identification of GSVA, PLAGE, single sample GSEA (ssGSEA) and combined z-score (zscore) on a leukemia data set. (A) Volcano plot of gene expression changes in the Leukemia data set. Genes highlighted in red form the first tercile of largest absolute fold changes, violet indicates the second tercile and blue the third tercile. (B-D) Adjusted rand index (ARI) indicating the accuracy of classifying the two groups of samples by hierarchical clustering of the enrichment scores produced by each of the compared methods at the top-5 differentially activated gene sets. The distribution of ARI values is formed by bootstrapping 1,000 times 10 samples from each sample group. Colors match the key given for genes in the volcano plot of (A) and show that, as expected, genes with larger fold changes lead to larger ARI values. However, when fold changes are small (B-C) and the underlying signature becomes extremely subtle, GSVA produces enrichment scores that lead to differentially activated gene sets which classify the two sample groups substantially better than using ssGSEA, zscore or PLAGE.

Back to article page