Skip to main content
Figure 3 | Genome Biology

Figure 3

From: Phylogenetically and spatially conserved word pairs associated with gene-expression changes in yeasts

Figure 3

Conserved word-pair templates are associated with significant changes in gene expression. (a) Total number of conserved word-pair template associations at different K-S significance values. The horizontal axis shows different FDR-corrected significance levels for the Kolmogorov-Smirnov test (see Materials and methods). The number of closely spaced word pairs meeting this cutoff for different minimum numbers of expression conditions is shown on the vertical axis. Word pairs were also filtered for an improvement of 10× over the K-S significance from any single word. (b) Relative enrichment for significant gene-expression associations compared to independently conserved words. Relative enrichment was computed as an odds ratio: the fraction of gene groups selected by conserved word-pair templates associated with significant gene-expression changes, divided by the fraction of gene groups selected by randomly conserved word pairs associated with significant gene-expression changes. Templates were chosen as the set of 989 word pairs showing dependent conservation and close spacing (χ2 > 31.1, spacing q < 0.05); the random word pairs included 624 pairs showing independent conservation and modest spacing constraints (χ2 < 31.1, spacing q < 0.15). The odds ratio is shown on the vertical axis; various FDR-corrected significance thresholds for gene-expression associations are shown on the horizontal axis. Word pairs were filtered for an improvement of 10× over the K-S significance from any single word. (c) Relative enrichment for significant gene-expression associations compared to co-conserved words that failed the close spacing test. Templates were chosen as the set of 989 word pairs showing dependent conservation and close spacing (χ2 > 31.1, spacing q < 0.05); the background word pairs included 2,737 pairs showing co-conservation, but no significant close spacing constraints (χ2 > 31.1, spacing q > 0.05). The odds ratio is shown on the vertical axis; various FDR-corrected significance thresholds for gene-expression associations are shown on the horizontal axis. Word pairs were filtered for an improvement of 10× over the K-S significance from any single word.

Back to article page