Structure-Based AI Small-Molecule Discovery

Research question and target

Compare structural evidence, experimentally measured binding values and predicted properties for therapeutic targets and previously reported compounds.

Internal target

Binding free-energy error of 1 kcal/mol or less

Programme

Discovery and data

Evidence on this record

Common synthetic checks

Evaluation data

• Official D3R Grand Challenge 4 Free Energy Sets

  2018 GC4 · publisher repository a7df7f521c234a03c97c26fe3755784381156197

  Manifest locked
• Direct thermodynamic measurements with independent scaffold groups

  Eligible reference set pending

  Blocked

Evaluation registry · Labels and split definitions (KR)

Sources and eligibility

• RCSB PDB

  Open

  The PDB archive distributes structure files under CC0. Comparable experimental binding free energies are unavailable for some structures.

• ChEMBL

  Open

  Activity and compound data are licensed CC BY-SA 3.0. Distinct activity endpoints and assay conditions require separate treatment.

• ZINC20

  Open paper · Data access unverified

  The authors describe public access. The data server returned an access error during review. Compound lists lack measured binding or toxicity evidence.

Boundary cases

• Evaluation leakage across molecular scaffolds and homologous target families
• Inconsistent assay conditions and units across binding measurements
• The research specification also sets limits of 0.4 structural similarity, 2.0 Å RMSD over 100 ns and five retrosynthetic steps

Common synthetic checks

• Relative binding free-energy format and error calculation · 8/8 matched
• Synthetic benchmark sample and scaffold separation · 3/3 matched

Expected and observed values for these cases

Review of the attached design

The code attached to this project was reviewed against its claims: submission format, units, blinded partition rules and comparison of binding values.

5-1 and 18-2 attachment review (KR)

What the official call asks

Led by a company that holds an AI drug-discovery platform, with a pharmaceutical or biotech partner. Outcomes include the platform and its efficacy results, a candidate ready for non-clinical entry with safety and efficacy test reports, and a substance patent filing.

Proposal requirements, printed pages 18–20 · All twelve requirements

Interpretation and next evaluation

The computational benchmark covers scores and chemical metrics. Patent novelty, synthesis, toxicity and therapeutic effects each require their own evidence.

Next check

Select records with measured binding values, specify assay conditions and lock the split by scaffold, target family and date.