Figure 1

Schematic overview of the method. The inputs are: a whole-transcriptome RNA-seq profile of a prokaryotic organism, the genome annotations and the corresponding map of operons collected by DOOR. The core process is represented by four steps. The first three steps determine a set of confirmed operons. While, in the last step the system trains and validates a NN-, a RF- and a SVM-based classifier on a list of confirmed OPs and NOPs. In output, these classifiers are used to reassess the operon structure annotated in DOOR. Further, a validation process is accomplished to computationally verify that there are not predicted regulatory signals between adjacent genes indicated as operon pairs.