Converter
Drop a vendor σ-profile file or a structure file. We turn it into a signed, schema-locked .mfsig.json. The conversion CLI is open and free; enrichment to Platinum-grade is per-molecule.
Supported source formats
8 vendors · auto-detectVendor file → open .mfsig.json. Atoms, geometry, σ-profile blocks, methodology, and a SHA-256 audit seal — extracted, normalised, signed. No login, no API key.
- ✓ 8 vendor formats auto-detected
- ✓ SHA-256 audit · 21 CFR Part 11 ready
- ✓ CLI · web UI · resumable batch
- ✓ Open spec, no vendor lock-in
Take a converted .mfsig further — add σ-profile from scratch, bias-corrected ΔG, direct DFT dipole, the automated 21 CFR Part 11 audit pack. Pick a grade by what your downstream pipeline needs to defend.
What is conversion · in three sentences
Detect the originating tool from the file's header or extension. Parse atoms, geometry, σ-segments, method block — whatever was persisted.
Move each field into the 5-pillar .mfsig schema. Re-canonicalise SMILES, derive the InChI-key, normalise units. Nothing recomputed, just reshaped.
SHA-256 the canonical payload, mint a UUID, stamp ISO timestamp + engine version. The output is a single tamper-evident file your auditors can replay forever.
Conversion is lossless within what the source format persists. What the vendor didn't write to disk (e.g. raw σ-segments from a CTD file) can't appear in the .mfsig — that's where enrichment kicks in.
CLI · production conversions
The open mfsig CLI is the real workhorse — same auto-detect, plus batch + resume + manifest.
# Single file (vendor auto-detected)
$ mfsig convert legacy_run.cosmo
✓ vendor detected: Turbomole
✓ wrote out.mfsig.json · SHA-256 e3b0c44298fc…
# Force a vendor when content sniff is ambiguous
$ mfsig convert cpcm.dat --vendor orca
# Convert back to a vendor format for legacy pipelines
$ mfsig convert aspirin.mfsig.json \
--to turbomole -o aspirin.cosmoBatch · 10⁶-file archives
Idempotent + resumable. Re-running on the same directory is a no-op. Manifest carries SHA-256 + provenance per drug.
$ mfsig convert /data/legacy_turbomole/ \
--batch --out /data/mfsig/
120 files · 7 vendors auto-detected · 0 errors
→ manifest.json with SHA-256 + provenance
→ re-run is a no-op (already-converted files skipped)
# Then enrich the bulk to Platinum (cohort API):
$ mfsig enrich /data/mfsig/ --tier platinumConvert → enrich → upgrade. The three-step flow.
git-for-σ-profile · v2.4+Conversion gives you a syntactically valid signed .mfsig from your legacy vendor file. Enrichment adds quality: an additional audit tier, chem-aware moments, or a grade upgrade (Pro → Platinum → Reference). Every step appends to the lineage chain — never rebuilds, never loses provenance.
Drop a legacy vendor file (Turbomole, COSMOtherm, ORCA, Gaussian, xtb, NwChem, MolForge CSV) → get back a signed .mfsig with the same physics data, in our schema.
Pick an additional audit tier (e.g. Compatibility, to feed a legacy COSMO-RS pipeline). Same molecule, second tier appended under physics_variants. Foundational work reused.
Promote Pro → Platinum (adds conformer ensemble + dft-relax geometry) or Platinum → Reference (TZVPD basis + expanded anchor validation). Charged on the delta, never the full price of a fresh Platinum / Reference run.
Every converted file is signed
21 CFR Part 11 · ALCOA-compatibleThe schema includes a SHA-256 of the canonical payload, a UUID v4, ISO 8601 UTC timestamp, and the engine version that produced the file. Tamper with a byte and verification breaks. Replay the conversion years later — same input bytes, same SHA-256.