Data mining in the discovery of novel anti-infectives
The strategies used in the discovery of anti-infectives is both similar and distinct to the challenges of discovering drugs for chronic human target conditions. It is clear that the properties of both the targets – in terms of mechanism of action, and often the physicochemical properties of the drugs themselves are distinct.
Previous comprehensive analyses of historical data have led to various strategies such as the prioritization of targets that have no orthologues in the human genome, or the selection of compounds that are more ‘antibacterial like’ for screening, in that they are larger and more polar, breaking the classic Lipinksi Rule of Five. These strategies have had, in aggregate, little impact in the discovery of novel validated antibacterial targets. It is therefore timely to review and challenge some of these assumptions that have typically shaped data-driven approaches.
Specifically, the presentation will attempt to address the following questions:
- Is antibacterial chemical space distinct from human-targetted drug space?
- How important is the lack of an orthologue in humans for successful targets?
- Are there informatics approaches to the discovery of novel combinations?
- How can AI and Data Science address resistance?
- Are we on the verge of realising the importance of cryptic pathogens in disease.