🤖 Confirmed, thanks — and it turns out to be two different problems, so this report was half right in a way that mattered.

Decarboxylation: the accession was wrong. UNIMOD:914 is Methylmalonylation on Ser (H(4) C(4) O(3), +100.016044). The record matching what MetaMorpheus actually searches — H(-2) C(-1) O(-1), −30.010565, on Asp and Glu, titled "Decarboxylation" — is UNIMOD:1915. The atom count in Mods.txt was correct all along, so no searched mass changes; what changes is the UNIMOD: accession written into .mzID output, which until now denoted a +100 Da modification on Ser while carrying a −30 Da mass on Asp/Glu.

Glu to PyroGlu: here the accession is right and the name is wrong. UNIMOD:28 is Gln->pyro-Glu ("Pyro-glu from Q"), H(-3) N(-1), −17.026549, Gln at any N-term — exactly the entry in Mods.txt (TG Q, N-terminal, H-3 N-1). UNIMOD:27 is Glu->pyro-Glu on Glu, with a different delta (H(-2) O(-1), −18.010565), so switching the accession to 27 would have put it at odds with both the mass and the motif. The label was the error — it named the wrong source residue — and it is renamed to Gln to PyroGlu.

The same audit turned up a second wrong accession that was not reported: Homoserine lactone on M (C-1 H-4 S-1, −48.003371) cited UNIMOD:10, Met->Hse (−29.992806), where the match is UNIMOD:11, Met->Hsl. That is the CNBr cleavage modification, so it is on a live path.

Both accession fixes are in smith-chem-wisc/mzLib#1245, together with a test that checks every hand-written Unimod accession in the shipped mod files against the bundled unimod.xml — 103 references audited, and those two were the only composition mismatches.

The fix lands in mzLib rather than here because the modification data files have moved into mzLib and are shipped in its NuGet package. It reaches MetaMorpheus only after an mzLib release, a PackageReference bump, and the pending change that removes MetaMorpheus's own copies of these files. Until then MetaMorpheus still ships EngineLayer/Mods/Mods.txt with DR Unimod; 914. in it, so this stays open.

Uh oh!

In PTM search assigned unimod id dosen't match real unimod id #2288

Description

Activity

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions