Resistance of bacteria to current antibiotics is an alarming health problem.
In this sense, Pseudomonas represents a genus of Gram-negative pathogens, which has
emerged as one of the most dangerous species causing nosocomial infections. Despite
the effort of the scientific community, drug resistant strains of bacteria belonging to
Pseudomonas spp. prevail. The high costs associated to drug discovery and the urgent
need for more efficient antimicrobial chemotherapies envisage the fact that computeraided
methods can rationalize several stages involved in the development of a new
drug. In this work, we introduce a chemoinformatic methodology devoted to the construction of a multitasking model for
quantitative-structure biological effect relationships (mtk-QSBER). The purpose of this model was to perform
simultaneous predictions of anti-Pseudomonas activities and ADMET (absorption, distribution, metabolism, elimination,
and toxicity) properties of organic compounds. The mtk-QSBER model was created from a large and heterogeneous
dataset (more than 54000 cases) and displayed accuracies higher than 90% in both training and prediction sets. In order to
demonstrate the applicability of our mtk-QSBER model, we used the investigational antibacterial drug delafloxacin as a
case of study, for which experimental results were recently reported. The predictions performed for many biological
effects of this drug exhibited a remarkable convergence with the experimental assays, confirming that our model can serve
as useful tool for virtual screening of potent and safer anti-Pseudomonas agents.
Keywords: Antibacterial, ADMET, delafloxacin, linear discriminant analysis, mtk-QSBER, Pseudomonas spp., TOMOCOMD.
Rights & PermissionsPrintExport