Fig. 3

Characterization and comparison of TSS clusters. a Cumulative percentage plot illustrating the fraction of Start-seq single nucleotide peaks within a given genomic distance of its nearest neighbor peak. Inset shows green region with expanded x-axis to illustrate that 50% of peaks are within 5 or fewer nt of the next nearest peak. b Histogram showing the distribution of peak cluster widths for obsTSSs identified in Start-seq data. Notably, 50% of peaks are narrower than 6 nt, in contrast with previously reported broader distributions of peak widths [13, 36]. c Boxplot describing the proportion of broad (left), peaked (middle), and weak (right) peak clusters across a range of clustering thresholds from our data (Red), Nechaev et al. Start-seq data [19] (Green), and Hoskins et al. CAGE data [13] (Blue). For additional details, see Additional file 1: Figure S3. d Scatterplot comparing shape index (SI) with pausing index (PI)