Figure 6.

Incorporation of conservation and word pairs provided more informative average expression profiles. Groups of genes whose TCRs contained various sequence features were summarized by the average of their gene-expression profiles. Each row in these diagrams represents a given gene group's average expression pattern under the conditions shown in each column: exposure to increasing concentrations of cadmium chloride (from 0.05 mM to 0.4 mM); 20 min after heat shock to 37°C (from 17°C, 21°C, 25°C, 29°C, and 33°C); an amino-acid starvation time course; a nitrogen-source depletion time course; progression into stationary phase (2 h, 4 h, 6 h, 8 h, 10 h, 12 h, 1 day, 2 days, 3 days, 5 days of growth) [30]. Representative conserved word-pair templates were chosen for analysis, corresponding to: (a) Msn2/4p-like sequences (CTCCCC and CCCTTA); (b) Cbf1p (CACGTG) and Met31/32p (TGTGGC) binding sites; (c) Mig1p (ACCCCA) and Ume6 (CCGCCG) binding sites; (d) Msn2/4p-like (CCCCTT) and Hsf1p-like (GAGAAA) sequences. For each of the panels (a-d) each row represents the average expression profile for gene groups chosen by different sequence features in their TCRs: single words found in S. cerevisiae (rows 1 and 2); single words conserved in three or more Saccharomyces genomes (rows 3 and 4); word pairs found in S. cerevisiae (row 5); word pairs conserved in three or more Saccharomyces genomes (row 6). Arrows above the columns indicate conditions under which gene groups sharing the conserved word-pair template (row 6) were significantly associated with gene-expression changes, at a p-value of 0.001 (K-S test after FDR correction for multiple testing).