is proud to announce
Excipient Compatibility Discovery Engine
A clinical-grade platform leveraging machine learning to predict API–excipient compatibility — replacing months of trial-and-error with confidence-scored results in seconds.
Finding compatible excipients for an Active Pharmaceutical Ingredient (API) traditionally requires extensive lab testing — weeks of work, significant cost, and no guarantee of success.
PharmDEX changes this. Input any API compound and our ensemble ML models score its compatibility against a curated library of 300+ pharmaceutical excipients. Every prediction comes with a calibrated confidence interval — transparent, reproducible, and research-ready.
Enter two SMILES strings — one API, one excipient — and get an instant ML-scored compatibility result with 2D molecular visualisation and confidence intervals.
Compare one API against up to 20 excipient candidates at once. Results ranked in a leaderboard, exportable as PDF or CSV.
Scan the entire master library against a single API. Get top-ranked excipients by ML confidence — ideal for early-stage research.
Submit new excipient compounds to the database. Every submission undergoes expert review before going live — ensuring data quality across the platform.
Generate research-ready reports with molecular structures, scores, and confidence intervals for publications and submissions.
Every single analysis produces a comprehensive suite of ML-driven visualisations — not just a score, but the full scientific reasoning behind it.
Explore API–excipient interactions in an interactive 3D space. Atom-level cross-attention weights are visualised as coloured bonds — rotate, zoom, and pan to inspect every interaction.
Colour-coded speed gauge showing the compatibility score with a clear COMPATIBLE / INCOMPATIBLE verdict.
Per-atom activity bars revealing which atoms interact most strongly and with which partner atoms.
Visual bridge diagram connecting API and excipient atoms, colour-coded by interaction strength — Strong, Moderate, or Weak.
8-axis spider chart comparing molecular properties — LogP, MW, TPSA, HBD/HBA, rotatable bonds, and more.
SwissADME-style 6-axis radar with an optimal drug-like zone overlay for both API and excipient profiles.
Horizontal bar chart showing property differences between API and excipient — instantly spot where they diverge.
26 known functional-group risk pair rules evaluated against your input — triggered rules highlighted in red.
Lipinski, Veber, and Egan rule checks with pass/fail status for every molecular property.
Enter the SMILES notation for your API. PharmDEX validates and renders the 2D molecular structure for confirmation.
Ensemble models evaluate molecular descriptors, interaction profiles, and physics-based features in sub-second inference.
Receive compatibility scores with calibrated confidence intervals, molecular structures, and research-ready export options.
PharmDEX is an initiative to make high-quality research tools accessible to all — completely free.
Accelerate formulation development with ML-driven predictions instead of months of trial-and-error experimentation.
Free for academic researchers and students. Generate citable, reproducible results for publications and theses.
Batch and deep-scan capabilities for high-throughput formulation screening at institutional scale.
Custom models with cross-attention mechanisms, GIN encoders, and 155+ physics-based molecular features.
A master library of 300+ pharmaceutical excipients, maintained through community contributions and expert review.
Sub-second predictions via optimised model serving, with batch processing support for multi-excipient workflows.
Conformal prediction sets quantify uncertainty. All results include confidence scores for reproducible outputs.
PharmDEX launches May 1, 2026. Visit Dr PP Research Lab to learn more about our work in computational pharmaceutical science.
Visit Dr PP Research Lab