By EVOBYTE Your partner in bioinformatics
Introduction: from atoms to files—how we encode protein structure
Proteins don’t just exist as sequences; they fold into intricate three‑dimensional shapes that determine what they do. To share those shapes, structural biologists publish coordinate files that software can parse, visualize, and analyze. Two formats dominate this space: the legacy PDB file and the modern PDBx/mmCIF format. Understanding what each one stores, how they differ, and where the community is heading will help you future‑proof your pipelines and avoid brittle parsing bugs.
PDB files: the legacy backbone of structural biology
A PDB file is a fixed‑width, column‑oriented text format born in the 1970s. It encodes atoms as ATOM/HETATM records and packs identifiers, coordinates, and metadata into strict 80‑column slots. That rigidity made PDB files easy to skim by eye and with line‑based tools, but it also created hard technical ceilings: single‑character chain IDs, five‑digit atom serial numbers, and other field limits. As structures grew—think ribosomes, viral capsids, and massive cryo‑EM assemblies—those ceilings turned into blockers. The format is now considered “legacy” and no longer extended with new content.
PDBx/mmCIF: a dictionary‑driven, scalable standard
PDBx/mmCIF (often just “mmCIF”) rethinks structure files as machine‑readable tables governed by a community dictionary. Instead of squeezing values into fixed columns, mmCIF names every data item, groups them into categories, and relates them explicitly—making referential integrity first‑class. Crucially, it removes the size caps: there’s no built‑in limit on the number of atoms, residues, or chains that a single entry can represent. The Worldwide Protein Data Bank (wwPDB) adopted PDBx/mmCIF as the standard archive format in 2014, and all processing and annotation across wwPDB partners now use it. Visualization and refinement tools have followed suit, so you can open mmCIF in mainstream applications without conversion.
What actually changes for your day‑to‑day analysis
If you’ve ever written a brittle PDB parser, mmCIF feels like a relief. Data are key–value or tabular, whitespace‑delimited, and governed by strict definitions in the PDBx/mmCIF Exchange Dictionary. That means fewer “special cases” and clearer mappings between sequences, chains, residues, atoms, and experimental metadata. On the flip side, some legacy tools still expect .pdb input, which is why you’ll sometimes see “best‑effort/minimal” PDB exports for very large structures. But the authoritative truth lives in mmCIF, and it carries richer metadata—citations, sample details, validation metrics—that you can surface in notebooks and automated reports.
Here’s a tiny Python example showing how to read either format with Biopython and count atoms. It’s intentionally boring—because file I/O should be.
from Bio.PDB import PDBParser, MMCIFParser
def load_structure(path):
if path.lower().endswith((".cif", ".mmcif")):
return MMCIFParser(QUIET=True).get_structure("model", path)
return PDBParser(QUIET=True).get_structure("model", path)
structure = load_structure("example.cif") # or "example.pdb"
atom_count = sum(1 for _ in structure.get_atoms())
print(f"Atoms: {atom_count}")
Why mmCIF matters for the future: IDs, size, and new methods
The trend lines are clear. Structures keep getting bigger and more complex, and hybrid methods integrate X‑ray, NMR, cryo‑EM, and more. The legacy PDB format simply can’t represent many of these entries without splitting or losing information. The wwPDB now distributes such large structures as single, complete mmCIF files; bundled legacy PDB versions, where they exist, are strictly “best‑effort.”
There’s also an important change on identifiers. The archive is projected to exhaust four‑character PDB IDs around 2028. After that, the wwPDB will issue extended, 12‑character accessions that are incompatible with legacy PDB files. In practical terms, future entries with new IDs will be available only in PDBx/mmCIF. If your code assumes four‑character IDs or only reads .pdb, now is the time to update it.
Summary / Takeaways
The PDB format built structural biology, but its fixed‑column design can’t keep up with today’s assemblies and tomorrow’s identifiers. PDBx/mmCIF, guided by a public dictionary, scales without artificial limits and captures richer metadata with clearer relationships. For analysts and developers, the best practice is simple: make mmCIF your default, keep conversions only for legacy tools, and validate your pipelines against dictionary definitions rather than column numbers. Doing that now will save you headaches when 12‑character PDB IDs arrive and when the next giant macromolecular complex lands in your queue.
Interested in more articles on bioinformatics formats?
Check for instance our previous post on File Format for Next Generation Sequencing