Abstract
The prediction of operons in Mycobacterium tuberculosis (MTB) is a first step toward understanding the regulatory network of this pathogen. Here we apply a statistical model using logistic regression to predict operons in MTB. As predictors, our model incorporates intergenic distance and the correlation of gene expression calculated for adjacent gene pairs from over 474 microarray experiments with MTB RNA. We validate our findings with known examples from the literature and experimentation. From this model, we rank each potential operon pair by the strength of evidence for cotranscription, choose a classification threshold with a true positive rate of over 90% at a false positive rate of 9.1%, and use it to construct an operon map for the MTB genome.