Preparation changes a molecular model. Decide what should change before applying automated repair, and retain the original file for comparison.
Define the downstream system
Write down the intended chains, construct boundaries, ligands, cofactors, catalytic ions, and relevant protonation conditions. Establish whether the next step is folding, docking, protein design, or molecular simulation.
A repair appropriate for an isolated protein may remove components essential to a protein-ligand calculation. A model with completed atoms is not automatically ready for every force field or docking engine.
Inspect the original structure
Use a viewer and inspect the file’s metadata. Check:
- The selected model and biological assembly.
- Chain identifiers, numbering, insertion codes, and sequence correspondence.
- Missing residues and heavy atoms.
- Alternate positions, nonstandard residues, and termini.
- Ligands, ions, waters, nucleic acids, and other partners.
Distinguish missing density from a known structural conformation. Reconstructing a loop introduces a hypothesis, not measured coordinates.
Configure PDBFixer deliberately
Open PDBFixer. Upload a compatible structure and review the current form. The service exposes controls for adding missing residues, adding missing heavy atoms, and handling nonstandard residues and other components.
Enable only the repairs appropriate for the planned calculation. Missing-residue reconstruction depends on available sequence information; do not assume that an incomplete coordinate file contains enough information to recover every segment.
Review hydrogen and solvent-related settings where exposed. Protonation and solvation choices should match the downstream model, rather than being chosen solely to make a file pass validation.
Preserve essential cofactors and chemical identities. Replacing a nonstandard residue or removing a ligand can change the scientific system even when the output remains syntactically valid.
Validate the repaired output
Compare the repaired file with the original:
- Confirm expected chain and residue identities and counts.
- Check that intended ligands, ions, and partners remain present.
- Inspect reconstructed regions for clashes or implausible geometry.
- Verify numbering after any conversion or editing.
- Confirm format and atom naming requirements for the next service.
If a workflow needs force-field parameterization, charge assignment, or specialized treatment of metals and covalent ligands, that work is separate from basic structure repair.
Keep a preparation record
Save the original, repaired output, PDBFixer job ID, enabled settings, and reasons for removing or changing components. Recheck hotspots and restraints against the repaired file before downstream submission.
For broader setup guidance, return to Preparing inputs. For folding rather than repair of an existing model, see Structure prediction.