AI-guided drug discovery workbench
Where AI meets the literature, the databases and the tools.
A workbench for structure-based drug discovery: form an idea, gather the evidence, run the analysis, and write it up — end to end, with every result still linked to where it came from. No code at any step.
✦ End to end, AI-assisted
- 1Find the papersSearch the literature, then let AI score every result against what actually matters to you — not just which words it matched.
- 2Collect and runPull structures, sequences and compounds from the major databases, then run a broad range of modern AI and computational tools on them — without leaving RxCubed.
- 3Write it upTurn what you found into a report — with every result still linked to what produced it.
✦ Always with you. Never without you.
An assistant that has read your project sits beside every screen. It answers, explains and drafts — and never acts on its own.
You: Which of the papers I saved actually measured binding, rather than citing someone else who did?
Assistant: Three of the eleven report a measured affinity of their own; the other eight cite those three. Want me to draft a note setting out the three measurements side by side? You'll see it before anything is saved.
An example exchange.
Keywords
Quality
- #3High-resolution structure of human carbonic anhydrase II complexed with acetazolamide0.94
- #1Carbonic anhydrase II: a review of its physiological roles0.41
- #4Clinical pharmacokinetics of acetazolamide in glaucoma0.35
- #2Sulfonamide derivatives as antimicrobial agents0.28
The paper ranked 3rd by keyword match scores highest against the quality.
Live Mol*, drag to rotate: KRAS, the protein called undruggable for forty years — until sotorasib.
Every source on the ring is integrated into RxCubed, not just linked to — whatever comes back is saved as evidence in the same store and provenance graph as everything else, one connected record instead of scattered accounts.
- RxCubed's tools and utilities — pocket detection, contact mapping, structure alignment, drug-likeness filters, format conversion and more, all running on the platform.
- RxCubed's connectors — RCSB PDB, AlphaFold DB, UniProt, PubChem, ChEMBL and InterPro, plus OpenAlex-backed literature search — searched and saved as evidence.
- 100+ Neurosnap tools (AI & Computational Biology) for structure prediction, design and docking, through your own Neurosnap account.
- Rowan's workflows (Molecular Modelling & Computational Chemistry) — pKa, tautomer search, bond dissociation, redox potential and FEP — through your own rowansci.com account.
Every result remembers what made it
A compound and a structure feed one tool; that tool's own output feeds the next. Every node here is a real result you could open, and the arrows are the exact lineage RxCubed keeps — the same graph you get for any run in a project.
- Start hereWhat RxCubed isThe workspace, its navigation, and how evidence and provenance are organised.
- New to the fieldDrug discovery basicsGenes to proteins, why shape decides what a drug can do, and which diseases this kind of work actually suits.
- How we workYou ask, we build itThis is how RxCubed gets built — tell us what would help, and we build it.
All of the documentation is free to read — no account needed.
DocumentationFree to start, and no credit card. A monthly allowance of computing time is included; heavier use is paid.