Inside Chemistry Space: design, ADMET, and explainable SAR

If Biology Space decides what to target, Chemistry Space decides what to make. It takes you from a bare SMILES string to a ranked series of candidates — designing molecules, predicting their properties, and, crucially, explaining what drives potency.

Design and predict

• Molecular design — Enumerate analogues around a scaffold and screen them in silico.
• ADMET prediction — Estimate absorption, distribution, metabolism, excretion, and toxicity up front.
• Retrosynthesis — Plan a synthetic route so a promising molecule is actually makeable.

Explainable SAR: the part we're proud of

A potency number tells you a molecule works. It doesn't tell you why — and without the why, the next molecule is a guess. Our SAR analysis attributes activity to specific substructures: feed it a congeneric series with activities, and it shades each atom and fragment by how much it lifts or drops potency.

"This nitro group at the para position is worth a full log unit of potency" is a design lead. "Compound 14 is potent" is not.

It also flags activity cliffs — near-identical molecules with large activity gaps — which are often the most informative pairs in a whole series.

A loop, not a report

The output feeds back into design. Learn which fragment carries potency, enumerate analogues around it, predict their ADMET, and send the survivors to the Work Station for binding validation. Each pass tightens the series.

Full input formats and worked examples live in the Chemistry Space docs.