Comparing docking scores
Assess input preparation and score comparability for molecular docking and candidate screening.
Skill version: 1.1.0. Resource: skill://nadoobio-docking-screening/references/scores.md.
Docking calculates possible molecular arrangements around a target. A pose is a predicted binding orientation. Scores come from a defined calculation and are not experimentally measured binding constants.
Read score units and preferred direction from the output contract. Do not merge scores from different engines, targets, or preprocessing into one ranking. Match target coordinates, binding regions, molecular states, and repeat conditions first.
Kd is the dissociation constant defined at binding equilibrium. Under the same conditions, lower values indicate stronger binding. Converting a computational score to Kd requires a physical definition and calibration evidence; matching units such as kcal/mol are insufficient.
Keep screening conditions and exclusion reasons with the results. High-ranking candidates are priorities for further experiments, not proof of binding, function, or clinical effect.
Materials to obtain before comparison
| Inputs and outputs | What to check | Next assessment |
|---|---|---|
| Target structure and binding region | Are the chain, structural state and search region the same? | Keep separate comparison groups if they differ. |
| Original candidate representation and prepared coordinates | Are charge, hydrogenation state, atom mapping and preparation method recorded? | Do not assume identical prepared inputs from identical original representations. |
| Poses and score table | Are the calculation method, version, score definition, units, preferred direction and repeat conditions the same? | Interpret rankings only within matching conditions and retain original scores. |
Hypothetical teaching example: scores of -8 and -10 from different methods cannot immediately be combined into one ranking, even if both use kcal/mol. Obtain calculation conditions and separate groups first. With the same method, target and preparation conditions, and a lower-is-better score, -10 may identify a candidate for earlier review within that group. It is not a measured binding constant or drug effect.
Check the reference used for coordinate comparisons. RMSD to the top computed pose differs from RMSD to an experimental reference pose. A column name alone does not establish reproduction of an experimental pose. Preserve predicted poses, scores and preparation records together so later ranking differences can be explained.
Sources: Binding and thermodynamics terminology, Virtual screening protocol connecting selection to follow-up validation. The protocol is a reference, not validation of these hypothetical numbers or a particular tool’s current input contract. Use the current result schema and method documentation for specific score definitions.