Pyteomics v5 compatibility: `Peptidoform` fails for user-defined modification names

[RalfG]

Peptidoform objects with non-UNIMOD/PSI-MS modification names cannot be constructed under pyteomics 5, blocking a pyteomics 5 compatibility upgrade.

pyteomics 5 changed proforma.parse() to eagerly resolve modification names against UNIMOD and PSI-MS databases during parsing. Any name not found in a database raises KeyError instead of falling back to a GenericModification as in v4.

To reproduce:

from psm_utils import Peptidoform

Peptidoform("AC[Carbamidomethyl]DEK")   # OK — name is in UNIMOD
Peptidoform("AC[Cmm]DEK")               # KeyError: 'Cmm'
Peptidoform("ACM[Oxidation (M)]EK")     # KeyError: 'Oxidation (M)'

Issues in practice:

• io.maxquant: MaxQuant uses names like Oxidation (M) and Acetyl (Protein N-term) that are not valid UNIMOD terms and cannot be parsed under pyteomics 5. Also see Peptidoforms parsed from MaxQuant output are not always valid #138.
• Any downstream tool passing user-defined or search-engine-specific modification names through Peptidoform will break.

For now, the Pyteomics dependency is capped to < 5 in psm_utils 1.5.5.