OperonSEQer: A set of machine-learning algorithms with threshold voting for detection of operon pairs using short-read RNA-sequencing data
Fig 4
Operon-SEQer is best used as an ensemble of methods and finds operons not annotated by the standard but detected by PACBIO SMRTseq.
(A) AUC scores of OperonSEQer operon pair calls on MicrobesOnline E.coli operons, and both OperonSEQer and MicrobesOnline operon pair calls on SMRT-seq E.coli operons. (B) Venn diagrams showing total number of operon calls on E.coli genes that pass threshold per method. (C and D) Recall (blue) and specificity (red) of the Operon-SEQer ensemble with algorithm agreement cutoffs of 1–6 for operon pair calls made by the standard (dotted lines), SMRTseq (dashed line), or by the union of calls made by both (solid line); (C) represents all available operon pair data for the new E. coli data sets and (D) represents operon pairs that have agreement between two or more replicates.