Moleculepath Bio runs structure-guided virtual screening for specialty CROs and mid-size biotechs — pharmacophore filtering, GPU docking, ADMET scoring, and Benchling-ready export in 3–5 business days.
From target structure submission to Benchling-ready hit short-list delivery.
Maximum compound library size processed through the full multi-stage pipeline per campaign.
Substructure alert filters applied to every short-listed candidate before delivery.
ChEMBL bioactivity and DMPK records used to train our ADMET re-ranking model.
Every campaign runs the same auditable stack: structure preparation, multi-stage docking, ML re-ranking, ADMET scoring, PAINS alerting, and structured data export. No ad hoc steps, no undocumented decisions.
Three sequential stages run without researcher intervention — pharmacophore filtering, Schrodinger Glide SP/XP docking, and ML re-ranking — reducing a 5M-compound library to a curated short-list of 20–50 candidates.
Submit a raw PDB file or AlphaFold2 output with binding-site residue coordinates. Automated protonation, water placement, and grid generation return a validated receptor ready for docking within four hours.
Every short-listed candidate is annotated with predicted CYP inhibition, hERG liability, aqueous solubility, oral bioavailability, and BBB penetration — drawn from a model trained on 1.2M ChEMBL and DMPK records.
Results ship as JSON and CSV pre-formatted for Benchling’s Molecule Registry and Results Table schemas. A single import step brings docking scores, binding-pose images, ADMET flags, and compound IDs into your existing registry.
All candidates are checked against 480 PAINS substructure filters from Baell and Holloway (2010) plus an extended in-house library of HTS frequent hitters. Alerts carry severity levels and handling notes for medicinal chemistry review.
Each campaign produces a structured report covering library attrition, docking score distributions, hit rate, and ADMET flag prevalence — available as a Benchling notebook export or PDF for regulatory and partner-review use.
The workflow is designed around what a small biotech team can actually hand off — a structure file, binding-site coordinates, and a compound library. Everything else runs on our end.
Upload a PDB or AlphaFold2 structure file along with binding-site residue coordinates and your compound library (up to 5M SMILES strings from internal or commercial sources). A brief intake form captures any binding-mode hypotheses or scaffold preferences.
Moleculepath prepares the receptor: protonation state assignment, binding-site water placement, conformer enumeration via OpenEye OMEGA, and Glide grid generation. The prepared receptor is returned to you for review before docking begins.
Pharmacophore filtering cuts the library by 80–90%. Glide SP docking runs on the filtered set, followed by XP docking on the top 5% of scorers. The ML re-ranker then incorporates ADMET predictions, synthetic accessibility, and PAINS alerts to produce the final ranked list.
A Benchling-native data package with 20–50 ranked candidates, docking scores, binding-pose images, ADMET flags, and synthesis-feasibility scores lands in your inbox within 3–5 business days. A campaign report PDF ships alongside for partner or regulatory review meetings.
The typical time from target structure to validated hit short-list at a specialty CRO using manual ligand docking campaigns — the problem Moleculepath was built to solve.
Share of synthesis batches that yield no binding activity above 10 µM in early-stage campaigns where ADMET and PAINS filtering were not applied before route-to-synthesis decisions.
Cheminformatics contractor spend per target campaign for mid-size biotech programs that lack in-house virtual screening infrastructure, before factoring in synthesis costs on false-positive hits.
Send us a target structure and binding-site coordinates. We’ll walk through what the pipeline would produce for your program and give you a timeline and scope estimate before any commitment.
