mfsig.com
Drug atlas · pre-computed coverage

Thousands of drug-like molecules, audit-graded and ready to screen.

The drug atlas is what coverage looks like when every entry is audit-grade. Signed .mfsig profiles for screening, lead-finding, and formulation candidate ranking — pre-graded so you don't pay per-molecule to know what's catalogue-covered.

Scope

Two cohorts, two purposes

The validation cohort proves accuracy on public anchors. The atlas cohort extends coverage across the drug-like space. Same audit grade on both — different question being answered.

84
Validation cohort · public anchor overlap
0.85
FreeSolv LOO MAE on validation cohort · kcal/mol
thousands
Atlas cohort · drug-like coverage
30+
Pharma-grade polymer carriers · ASD-relevant

Validation is the page that proves accuracy. Atlas is the page that gets work done. Both ship as signed .mfsig files at the same audit grade — what differs is the question, not the engineering.

What you get

Pre-computed χ-matrix · drug × polymer · drug × solvent

The atlas is a sparse-matrix of pre-computed pair scores: every atlas drug × every catalogue polymer χ, plus every atlas drug × the top-200 reference solvents.

tier — Pro

Single-drug lookups

Pay-per-molecule access: request one drug, get the row of the χ-matrix for that drug against all catalogue polymers and reference solvents.

tier — Platinum

Full matrix access

Subscription: get the full χ-matrix as a signed cohort.tar.gz. Monthly refresh as the atlas grows. Use for portfolio-wide screening.

tier — Reference

Enterprise + custom

Custom drug lists, custom polymer panels, custom solvent baskets. SLA-backed. White-glove cohort generation against your in-house chemistry.

What it's for

Five concrete workflows

The atlas is most useful when you need coverage you don't want to pay per-molecule for. Five workflows our customers run on it.

workflow 1

Formulation candidate ranking

For a target API, rank the catalogue polymers by predicted χ. Top-K hits are your first-pass ASD formulation candidates.

workflow 2

Solvent screen for a target drug

For a target API, rank 200 reference solvents by Henry's constant or activity coefficient. Identify the best crystallisation / dissolution solvent.

workflow 3

Patent-lead generation

Identify ternary rescue compositions (drug + polymer A + polymer B) where binary χ fails but ternary works. Patentable claim space.

workflow 4

Portfolio risk assessment

For a pipeline of N drug candidates, quickly identify which are likely brick-dust (BCS class II / IV) and need formulation intervention.

workflow 5

Quality control on partner deliverables

When a CDMO submits a formulation, cross-check against the atlas: do their χ values look reasonable for that drug-polymer pair? Catch outliers early.

Access

The atlas is reproducible

Every mfsig is recipe SHA-pinned + vendor SHA-validated. The full atlas of 6 876 mfsigs is regeneratable from raw data in ~2 days, <$50.