PDBx/mmCIF structure support

[RubenChM]

Internal implementation for the structure already supports reading PDBx/mmCIF, but some downstream tasks that accepts a file as input may not.

Current structure file use legacy pdb format which comes with some shortcomings like:

• Number of atoms limited to 999999
• Number of residues limited to 9999
• Number of chains limited to 52
• Name of residues to 4 letters
  • gro files can have bigger residue names like in A01J4

Useful links:

1. Understand:
   1. 101.
   2. Guide.
   3. Ecosystem.
2. Software.
   1. https://gemmi.readthedocs.io/en/latest/cif.html#reading-and-writing

[RubenChM]

• Dependency graph visualization:

• Generated compatibility report:

Reader/program	Direct tasks	PDBx/mmCIF support	Current workflow result
PyTraj/CPPTRAJ	frames, average, perres, pairwise, dist, hbonds, density	Yes	frames and average worked directly in smoke tests. The remaining tasks are blocked by the workflow’s PDB-only parsed Structure. PyTraj documentation
MDTraj	pca, clusters, dihedrals, markov, custom-frame screenshot	Yes	Native loading works. pca, clusters, and dihedrals nevertheless fail when trajectory reduction is needed because that helper uses GROMACS. Most also require the PDB-only internal structure. MDTraj source
VMD	inter	Yes, through its PDBx plugin	Native CIF+XTC loading worked, but inter also needs the internal structure and sometimes the GROMACS reducer. Long chain names may be truncated by this VMD build. VMD PDBx plugin
GROMACS	firstframe, rgyr, pmeta; trajectory reduction used by several other tasks	No	GROMACS 2025.4 rejected .cif for -s; accepted extensions are TPR/GRO/G96/PDB/BRK/ENT. GROMACS trjconv
MDAnalysis	mda_univ	No	MDAnalysis 2.10 raises “CIF isn’t a valid topology format.” This also blocks rmsf, thickness, apl, lorder, linter, channels, inchikeys, and memmap. MDAnalysis topology formats
mdpocket	pockets	No	The installed fpocket supports mmCIF, but the workflow invokes mdpocket, whose trajectory mode still requires a PDB topology. Official fpocket/mdpocket guide
CPPTRAJ + Curves+	helical	Mixed	CPPTRAJ successfully read CIF, but Curves+ documents only AMBER or PDB topology input, so the complete task is incompatible. Curves+ manual
BioBB/FATSLiM	apl	No in this path	The wrapper accepts TPR/GRO/G96/PDB-like inputs and first loads them through MDAnalysis, so CIF fails before FATSLiM runs. BioBB MEM documentation

• Blockers:
  • GROMACS: support is being currently being implemented.
  • MDAnalysis: support is being currently being implemented.
  • Curves+: we host the source code at https://github.com/mmb-irb/Curves_plus/tree/master. Maybe fork it or update to add a gemmi reader.
  • mdpocket: fpocketalready has open_mmcif/read_mmcif but there is no ongoing effort.
  • apl: should support once we change the current implementation to LiPyphilic with the update MDAnalysis. Also Add neighbor selection functionality to AreaPerLipid analysis p-j-smith/lipyphilic#164