Fig. 4

sRNA ORFs with evidence for translation are preferentially predicted to be protein-coding. Ribosome profiling data from [25] was mapped to B. subtilis 168 and E. coli K12. sRNA ORFs were annotated as translated based only on ribosome profiling signal within three codons of the start or stop codon. a sRNA ORFs were separated into those independent of annotated ORFs and those antisense to annotated ORFs. The fraction of ORFs translated (black bar) is compared to mock ORFs matched for length and overlap properties (white bars with 95% confidence intervals). b The predicted coding probability of translated sRNA ORFs is compared to non-translated sRNA ORFs in a histogram for B. subtilis (top) and E. coli (bottom). c Top: B. subtilis sRNA sbsu2300.1 has three potential coding ORFs, but only one is predicted to be coding. The start and stop codons of this ORF correspond to ribosome profiling peaks, while the ORFs predicted as noncoding do not. Bottom: E. coli sRNA seco4050.1 (CsrC) has four ORFs, two of which are predicted as coding and overlap with ribosome profiling peaks. The coding probability can help distinguish between the coding frame for overlapping ORFs, as in the first two in this sRNA