Rapid feasibility assessment of virtual hits

When triaging virtual hits from docking, generative AI, or fragment expansion, the constraint isn't binding affinity, it's whether you can make the molecule before the project clock runs out. This tutorial uses route feasibility as a fast triage filter.

Steps

1. Pull the top 50-100 virtual hits from your scoring pipeline.
2. Open Submissions → Library Synthesis, Quick List mode, and paste the hit SMILES list.
3. Before submitting, turn on Commercial Stock Filter. This restricts the search to routes whose starting materials are commercially purchasable, the strongest single feasibility signal you can apply at submission.
4. Set route length to 4 (shortest plausible for moderately complex hits). You can rerun with longer routes for any hits that come back empty.
5. Submit. Wait for the library to finish.
6. Filter triage:
   • Empty result = no commercial-stock route found at length 4. Park or rerun with relaxed constraints.
   • High confidence + short route + cheap starting materials = promote to wet-lab queue.
   • High confidence + commercially available but long route = depends on project timeline.

Output

A small set (typically 10-20% of the original hits) that survives all three feasibility filters. Run those.

The hits that don't survive triage often resurface as failed lead optimization candidates a quarter later. Cutting them now saves real calendar time.