PDB Search FAQ

Q: How do I choose the best structure for my analysis?

Prioritize these factors in order: Resolution: Lower is better. Aim for <2.5 A for docking work. Bound ligand: For docking studies, prefer holo structures (with a co-crystallized ligand) over apo structures. The ligand defines the active binding conformation. Completeness: Check for missing residues, especially in loops near the binding site. Organism: Match your target organism when possible. Cross-species structures may have active site differences. Recency: Newer depositions may benefit from improved experimental techniques and refinement methods.

Q: What is the difference between PDB and mmCIF format?

PDB format is the legacy text-based coordinate format used since the 1970s. It uses fixed-width columns and has limitations including a maximum of 99,999 atoms and 62 chains. mmCIF (Macromolecular Crystallographic Information File) is the modern replacement format adopted as the standard for new PDB depositions. It has no atom count limits, uses a more flexible key-value structure, includes richer experimental metadata, and is machine-parseable. Most modern molecular viewers and analysis tools support both formats. When in doubt, PDB format has the broadest compatibility.

What resolution should you look for? Below 2.0 Å for docking. Plus PDB versus mmCIF, apo versus holo, and why one protein has many depositions.

These answers cover what PDB Search retrieves, how to judge a structure by resolution and bound ligands, and how PDB and mmCIF formats differ — check here when you're choosing the best structure for docking or downstream analysis.

Frequently asked questions

What is PDB Search and what does it retrieve?

PDB Search queries the RCSB Protein Data Bank, the global archive of over 220,000 experimentally determined three-dimensional structures of biological macromolecules. It retrieves structures determined by X-ray crystallography (~85% of entries), cryo-electron microscopy (~12%), NMR spectroscopy (~2%), and other methods. Results include metadata such as resolution, experimental method, organism, release date, and links to view structures on the RCSB website.

How is PDB Search different from ESMFold?

PDB Search retrieves experimentally determined structures from the RCSB database. These are real, measured atomic coordinates derived from laboratory experiments. ESMFold predicts structures from amino acid sequences using an AI language model (ESM-2). Use PDB Search when an experimental structure exists for your protein of interest, especially when you need structures with bound ligands for docking studies. Use ESMFold when no experimental structure is available, for novel or engineered sequences, or when you need a quick prediction.

What resolution should I look for?

Resolution is the primary quality indicator for X-ray crystal structures:

Below 2.0 A: High resolution. Excellent for docking, binding site analysis, and detailed molecular modeling.
2.0-2.5 A: Acceptable for most structure-based drug design work. Side chain positions are generally reliable.
2.5-3.0 A: Moderate. Backbone is reliable, but side chain rotamers may be uncertain. Use with caution for detailed binding analysis.
Above 3.0 A: Low resolution. Useful for overall fold characterization but not recommended for docking or binding site analysis. NMR structures do not have a resolution metric and are assessed by different quality criteria.

Can I download structures in different formats?

Yes. PDB Download supports two formats: PDB format (the legacy text-based format, widely compatible with most molecular viewers and analysis tools) and mmCIF format (the modern replacement with no atom count limits and richer metadata). PDB format is the default. Use mmCIF for very large structures (>99,999 atoms) such as ribosomes or viral capsids, or when you need complete experimental metadata.

Can downloaded structures be used for docking?

Yes. PDB structures feed directly into DiffDock for molecular docking. DiffDock also accepts a PDB ID directly, so you can skip the download step entirely. Structures can also feed into P2Rank for binding pocket prediction, PDB2PQR for protonation state assignment and electrostatics, and MM-GBSA for binding free energy rescoring.

Why are there multiple structures for the same protein?

Proteins are frequently crystallized under different conditions (pH, temperature, buffer), with different bound ligands (drugs, substrates, cofactors), at different resolutions, or captured in different conformational states. Each PDB deposition represents a unique experiment. Having multiple structures provides complementary information about the protein's flexibility, binding modes, and conformational landscape. For drug discovery, structures with different bound inhibitors are particularly valuable.

How do I choose the best structure for my analysis?

Prioritize these factors in order:

Resolution: Lower is better. Aim for <2.5 A for docking work.
Bound ligand: For docking studies, prefer holo structures (with a co-crystallized ligand) over apo structures. The ligand defines the active binding conformation.
Completeness: Check for missing residues, especially in loops near the binding site.
Organism: Match your target organism when possible. Cross-species structures may have active site differences.
Recency: Newer depositions may benefit from improved experimental techniques and refinement methods.

What is the difference between PDB and mmCIF format?

PDB format is the legacy text-based coordinate format used since the 1970s. It uses fixed-width columns and has limitations including a maximum of 99,999 atoms and 62 chains. mmCIF (Macromolecular Crystallographic Information File) is the modern replacement format adopted as the standard for new PDB depositions. It has no atom count limits, uses a more flexible key-value structure, includes richer experimental metadata, and is machine-parseable. Most modern molecular viewers and analysis tools support both formats. When in doubt, PDB format has the broadest compatibility.