Fig. 1

Genome-wide analysis of TSS selection in S. cerevisiae. a Overview of method and description of simple metrics used in analyzing TSS distributions at yeast promoters. b Reproducibility of TSS-seq analysis demonstrated by heat scatter correlation plots determine RNA 5′ ends across all genome positions with ≥ 3 reads in each library for biological replicates of WT, rpb1 E1103G, and rpb1 H1085Y libraries. Colors indicate the plot density from cool to warm (low to high estimated kernel density). Pearson r is shown. c Heat map illustrating hierarchical clustering of Pearson correlation coefficients between aggregate (combined biological replicate) libraries for all strains. Clustering illustrates increased correlation among known reduced function rpb1 alleles (“slow” or LOF) are increased correlation among increased activity rpb1 alleles (“fast” or GOF). WT shows intermediate correlations with both classes. d Core promoters (n = 6044) predicted by Rhee and Pugh from GTF ChIP-exo data were used to initially map TSSs. TSSs were row normalized to illustrate the distribution within each window; TSSs generally map downstream of predicted core promoters for most but not all promoter windows. Note that the resolution of the figure will have less pixels that promoter rows (6044). e Determination of change in median TSS position (upstream shift in median position is negative (cyan), downstream shift in median position is positive (orange), see the “Methods” section) for promoters with ≥ 200 reads (n = 3494). Heat map shows individual yeast promoter regions hierarchically clustered on the y-axis with the measured TSS shift for hierarchically clustered TSS usage affecting mutants on the x-axis