Target-to-Lead In Silico · Cambridge, MA
Five IND-ready candidates. Not five thousand virtual hits.
Manas AI runs target-to-lead in silico — predicting binding affinity, ADMET profiles and off-target liabilities before a molecule enters a wet lab. Pharma-grade computational chemistry, built to replace screening cascades.
The Bottleneck
Wet labs synthesize what computers should already have filtered.
Standard virtual screening hands pharma teams thousands of candidate structures ranked by shallow docking scores. The real filters — metabolic stability, hERG liability, CYP inhibition, BBB penetration — only appear at bench cost and 6-month delay. We run those filters computationally, before synthesis, at the scale of millions of conformations.
Our Approach
Three interlocking models, one ranked output.
Graph neural network trained on 14M+ protein-ligand complexes from PDB, ChEMBL, and proprietary assay data. Reports ΔG estimates with uncertainty quantification — not just rank order.
Ensemble of QSAR models covering 48 endpoints: metabolic stability (CYP1A2, CYP3A4, CYP2D6), plasma protein binding, aqueous solubility, hERG block, P-gp efflux, blood-brain barrier permeation, and hepatotoxicity flags.
Pan-proteome docking against 2,300 off-target structures. Flags promiscuity risk before synthesis. Reduces late-stage attrition from selectivity failures — historically responsible for 22% of Phase II discontinuations.
Pipeline
Current candidates
Five programs in progress. Three shown publicly.
Process
We take your validated target. We return a ranked candidate list with full in silico dossier.
You provide the crystal structure, homology model, or cryo-EM density. We accept PDB, mmCIF, or raw coordinate data.
Our pipeline screens your compound collection or generates de novo structures using fragment-based growth. Typically 10M–100M conformations evaluated in 48–72 hours.
You receive a ranked hit list with ΔG estimates, ADMET flags, off-target liabilities, synthetic accessibility scores, and a written rationale for the top 20 candidates.
Talk to us about your target.
We work with pharma R&D teams and biotech discovery groups. Initial consultation is a 45-minute call — bring your target protein and we can scope a campaign on the spot.