Neglected diseases are infections that thrive mainly among underdeveloped
countries, particularly those belonging to regions found in Asia, Africa, and America.
One of the most complex diseases is noma, a dangerous health condition characterized
by a polymicrobial and opportunistic nature. The search for potent and safer
antibacterial agents against this disease is therefore a goal of particular interest.
Chemoinformatics can be used to rationalize the discovery of drug candidates,
diminishing time and financial resources. However, in the case of noma, there is no in
silico model available for its use in the discovery of efficacious antibacterial agents. This work is devoted to report the
first mtk-QSBER model, which integrates dissimilar kinds of chemical and biological data. The model was generated with
the aim of simultaneously predicting activity against bacteria present in noma, and ADMET (absorption, distribution,
metabolism, elimination, toxicity) parameters. The mtk-QSBER model was constructed by employing a large and
heterogeneous dataset of chemicals and displayed accuracies higher than 90% in both training and prediction sets. We
confirmed the practical applicability of the model by predicting multiple profiles of the investigational antibacterial drug
delafloxacin, and the predictions converged with the experimental reports. To date, this is the first model focused on the
virtual search for desirable anti-noma agents.