Inferring functionally related genes in microbial genomes is an important problem, which has been addressed by various gene cluster detection methods. Existing models well capture genomic inversions, gene duplications and insertions across a pair of genomic regions. Clusters involving multiple regions are indirectly inferred through constituent pairs. In this paper, we improve upon current work by a) formally modeling gene losses, and b) directly inferring gene clusters comprising of multiple genomic regions. Our preliminary results show a better sensitivity and specificity in identifying verified operons when compared with HomologyTeams model.
展开▼