How to Write the Methods Section for a Molecular Docking Study

A molecular docking methods section must let another researcher reproduce your run without contacting you. State the structure source and PDB ID, the software and version, how you prepared the receptor and ligand, the grid box center and size, the scoring function and search settings, and your redocking validation with its RMSD. If any one of those is missing, the study is not reproducible.

Most rejected docking manuscripts are not rejected because the science is wrong. They are rejected because the methods section leaves a reviewer unable to repeat the work. This guide gives you a reporting checklist, a worked example you can adapt to your own target, and the specific numbers reviewers look for. If you are new to docking mechanics, read our molecular docking pillar guide first, then come back to write it up.

What must a molecular docking methods section include?

A complete docking methods section answers six questions in order: where the structures came from, how you prepared them, where you searched, how you scored poses, how you validated the protocol, and what software produced the numbers. Reproducibility is the single standard. A reader with your paper and public databases should be able to arrive at the same result.

The reason this matters goes beyond passing review. Docking scoring functions are approximations, so results depend heavily on choices that are invisible in a final figure: the protonation state you assigned, where you centered the box, how many poses you sampled. Two labs docking the same ligand into the same protein can report different affinities purely because their preparation differed. Reporting those choices is what separates a result a reader can trust from a number they cannot check.

What are the six blocks every docking methods section needs?

• Structure source: the PDB ID, the experimental method and resolution, and whether you used a co-crystal, an apo structure, or a homology model.
• System preparation: removal of waters and additives, added hydrogens, assigned charges, protonation states, and the software used for each step.
• Search space: the grid box center coordinates and its dimensions, in Angstroms.
• Docking and scoring: the program and version, the scoring function, exhaustiveness or equivalent sampling depth, and the number of poses kept.
• Validation: redocking of a known ligand and the resulting RMSD against the native pose.
• Analysis: how you selected the final pose and which tools produced interaction diagrams or figures.

Why do reviewers reject docking methods sections?

Reviewers reject on omission, not elaborate error. The most common gap is a missing or vague grid box. “The binding site was targeted” tells a reader nothing, because the box center and size control the entire search and change the result. The second most common gap is no validation: a study that never redocks a known ligand offers no evidence its protocol recovers real binding modes. The third is a missing software version, which matters because scoring can change between releases.

Named, verifiable specifics are what make the difference. The RCSB Protein Data Bank holds more than 200,000 experimentally determined structures, so citing an exact PDB ID with its resolution is trivial and expected. When a reviewer sees a bare protein name instead of an ID, that is the first sign the methods were not written to be reproduced.

What exactly do you report for the structure and its preparation?

Report the PDB ID, the experimental technique (X-ray, cryo-EM, or NMR), and the resolution. State that you removed crystallographic waters and buffer additives, and name any cofactors or metals you kept, because those often sit in the active site and change docking. For the receptor, report that you added polar hydrogens and assigned partial charges, and name the charge model. AutoDockTools assigns Gasteiger charges and merges non-polar hydrogens by default, as described by Morris and colleagues (2009) in the Journal of Computational Chemistry, and stating this explicitly saves a reviewer from guessing.

For the ligand, report its source (a database ID such as a PubChem CID or DrugBank ID, or a drawn structure), how you generated the 3D conformer, the protonation state at your chosen pH, and the charge assignment. If you set specific rotatable bonds as flexible, say so. Our step-by-step guide on preparing a protein and ligand for docking covers each of these preparation steps in full, and every choice you make there is something the methods section must record.

How do you report the grid box (search space)?

Report three center coordinates and three box dimensions, all in Angstroms, plus how you chose them. If you centered on a co-crystallized ligand, say that and give the coordinates. If you used a predicted or literature-defined pocket, name the method. A box large enough to cover the pocket with a small margin, but no larger, is the goal, because an oversized box lets the search wander and weakens the result. Our guide on setting the grid box center and size in AutoDock Vina explains how to pick these values; the methods section simply records the numbers you used.

A reader should be able to paste your reported center and size into a Vina configuration file and reconstruct your exact search space. That is the test to apply before you submit.

How do you report the docking run and scoring?

Name the program and its exact version (for example, AutoDock Vina 1.2.0), the scoring function, and the sampling depth. In Vina the relevant setting is exhaustiveness, whose default value is 8; if you raised it, report the value and, for full reproducibility, the random seed. State how many poses you generated and how you selected the one you analyzed, usually the top-ranked pose by predicted binding affinity, reported in kcal/mol. If you are unsure how to read those scores, our guide on interpreting docking results and binding affinity explains what the numbers mean.

Cite the docking engine’s primary paper. AutoDock Vina should be cited as Trott and Olson (2010), and the current engine as Eberhardt, Santos-Martins, Tillack and Forli (2021), “AutoDock Vina 1.2.0”, in the Journal of Chemical Information and Modeling. The original authors describe the tool plainly: AutoDock Vina “significantly improves the average accuracy of the binding mode predictions compared to AutoDock 4.” Quoting the source, rather than overstating what docking can do, is exactly the tone a reviewer wants.

How do you report validation (redocking and RMSD)?

Validation is the block students most often skip and reviewers most often demand. Report that you redocked a co-crystallized ligand into its receptor using the identical protocol, and give the RMSD between the docked pose and the native crystal pose. The long-standing convention is that an RMSD below 2 Angstrom counts as a successful reproduction of the known binding mode. Name the tool you used to measure it, and prefer a symmetry-aware method so chemically equivalent atoms are not counted as errors. Our walk-through on validating a docking protocol by redocking and RMSD covers this end to end.

One sentence of validation changes how the whole study reads. It lets you write that your exact settings recover an experimentally determined pose, which converts your predictions for untested ligands from guesses into defensible results.

What does a complete example methods section look like?

Here is a worked example you can adapt. It uses a real, widely studied complex: PDB entry 1HSG, HIV-1 protease bound to the inhibitor indinavir (ligand code MK1), solved by X-ray crystallography at 2.0 Angstrom resolution. Replace the target, IDs, and numbers with your own.

“The crystal structure of HIV-1 protease in complex with indinavir was retrieved from the RCSB Protein Data Bank (PDB ID: 1HSG, X-ray, 2.0 Angstrom resolution). Crystallographic waters and the co-crystallized ligand were removed from the receptor. Polar hydrogens were added and Gasteiger partial charges were assigned in AutoDockTools 1.5.7, and the receptor was saved in PDBQT format. Ligand structures were retrieved from PubChem, protonated at pH 7.4, and prepared with the same charge model. Docking was performed with AutoDock Vina 1.2.0. The search box was centered on the coordinates of the co-crystallized ligand and sized to enclose the binding pocket with a margin (20 x 20 x 20 Angstrom). Exhaustiveness was set to 16 and the nine top-ranked poses were retained. The pose with the most favorable predicted binding affinity (kcal/mol) was selected for analysis. The protocol was validated by redocking indinavir into the prepared receptor; the top pose reproduced the crystallographic binding mode with an RMSD below 2 Angstrom, measured with a symmetry-corrected tool. Protein-ligand interactions were visualized in PyMOL.”

That paragraph names every choice a reader needs. It cites a real structure, states the versions, gives the box, reports validation, and never claims more than docking can deliver.

Reproducible versus rejected: what changes line by line

Written by the StemSkills Lab team, whose members have more than a decade of combined experience in sequence and structural bioinformatics, drug discovery and design, and multiscale molecular modeling. For where this fits in a full computational-biology skill set, see our computational biology skills roadmap.

Frequently asked questions

How long should a docking methods section be?

Long enough to be reproducible, usually two to four paragraphs. Length is not the target; completeness is. If a reader can rebuild your input files and search space from what you wrote, the section is the right length.

Do I need to report the software version?

Yes. Scoring behavior and defaults can change between releases, so a bare tool name is not reproducible. Report the exact version, for example AutoDock Vina 1.2.0, and cite its primary publication.

What if I used a homology model instead of a crystal structure?

Report the template PDB ID or IDs, the modeling software, and how you assessed model quality. A homology model is acceptable, but the reader must be able to see which structure it came from and how reliable it is.

Is redocking validation really required for a thesis?

For any docking study with a co-crystallized reference available, yes, and most examiners expect it. It is the strongest single piece of evidence that your protocol recovers real binding modes rather than plausible-looking artifacts.

Should I report binding affinity as a precise or approximate value?

Report the predicted value in kcal/mol as the software returns it, but describe it as a docking score, not an experimental measurement. Docking ranks and estimates; it does not measure affinity, and reviewers expect that distinction in your wording.