During the last three months I have utilised a combination of KNIME, SeeSAR and FastGrow to produce a library of molecules based on Evybactin. My first task was designing a workflow that could process commercial fragment libraries for growing. This was my first hurdle as I had to decide an approach that would give synthetically possible outcomes, while not limiting the substrates used. My solution was to create a chemistry driven workflow that processed fragments based on their functionality, and the chemistry possible with it – hoping to avoid creating strange linkages that couldn’t be synthetically replicated. My processed libraries (50k+ compounds from Enamine/Maybridge) were then used with FastGrowDBcreator and FastGrow to produce over 6k novel molecules. Using SeeSAR I then triaged the compounds, deciding on around 15 molecules that would be synthesised and tested against M. tb gyrase. Synthesis is ongoing but is yielding good results, with some molecules sent for screening.
After 3 months, Harry has achieved the following milestones:
- This milestone was achieved by taking a selection of diverse commercial fragment libraries and processing them into growth substrates. Designation of their respective linker atoms was done using a KNIME workflow that filtered the input for specific structural motifs, like amines or hydroxy groups. These functional groups were then transformed into linkage motifs that represented their structure after undergoing a typical toolbox reaction, for instance amide or metal-catalysed coupling. The outputs of this processing were then used to grow from a range of indole templates docked in SeeSAR, to ensure the best orientation and growth vector. These templates represented the other substrate from the typical toolbox reactions. The aim of this workflow was to use a chemistry-driven approach to ensure the synthesis of the outputs was more accessible. This milestone was finally reached after using the FastGrow module to produce over 6k novel molecules.
- Achieving this milestone started with the triaging and analysis of the diverse grown fragments. Despite designing a workflow that made synthesis as easy as possible, by utilising 1-step reactions from commercially available reagents, certain scaffolds were much more complicated than others. For a project with more funding this would have been no issue, simply buy the fragments. In this case it wasn’t financially feasible and more affordable starting materials would need to be used. A shortlist was created from estimated binding affinities generated by SeeSAR and overall structural similarity, to avoid synthesising scaffolds with similar 3D-shapes. 15 compounds were then selected based on the ease of their synthesis. This process is still ongoing with one molecule made and sent for testing, with several others in progress. Given the range of functionality, conditions cannot necessarily be used for analogous reactions, this setback is making some targets difficult to reach than others.
- This milestone is yet to be achieved as it requires the successful synthesis of the grown-fragment compounds and activity against M. tb gyrase. My hope is to use the biological data to aid expansion of SAR, screening and replacing motifs using SeeSAR’s re-coring tool as well as growing further from promising compounds with a wider range of fragments (using in-house and additional commercial libraries). This kind of computational approach to SAR expansion will hopefully result in increasingly diverse scaffolds, steering away from the typical process that results in very similar compounds to that already in the literature. A goal of this milestone is also to assess the importance of the Evybactin-based core of the indole, validating it and searching for potential replacements. Biological testing is specifically referring to M. tb DNA gyrase supercoiling assays that assess a compound’s ability to inhibit the enzyme. It is these values that will be used to inform further design efforts.