The paucity of new antibacterial agents in clinical development for the treatment of resistant Gram-negative infections is well known. Less than a handful of these agents possess a novel mechanism of action; the vast majority belong to existing antibiotic classes.
A major factor that has led to this inadequate clinical pipeline has been disengagement from antibiotic research by drug companies for commercial reasons. Another is the technical difficulty, especially against Gram-negatives; several major Pharma have reported analyses of their historical HTS campaigns indicating very limited success. One interpretation of these findings is that antibiotics fall into a different area of physicochemical property space from “normal” drugs. This is a reasonable hypothesis, given the complex cell envelope of Gram-negative organisms and many antibacterial drug discovery projects are unsuccessful not because they fail to find a potent inhibitor of their molecular target, but because of failure to accumulate at the target site sufficiently to have an antibacterial effect.
In this presentation, I will briefly review the work directed to identifying “accumulation rules” and how we at Oxford Drug Design are working to extend them using a combination of traditional cheminformatics approaches coupled with modern machine learning methods, illustrated using examples from one of our internal antibacterial drug discovery programmes.
