The reference score and the margin

The number to beat is not taken from a paper. It is measured here, on the same receptor file and in the same box a submission will be docked into, by re-docking the drug that was crystallised in that pocket.

Why re-dock it

A published affinity comes from an experiment, in different units, on different protein preparations. Comparing a Vina score to it would compare two different things. Re-docking the crystal ligand puts the reference and every submission through exactly the same program, the same settings and the same file, so one number decides and nothing else does.

Five runs, one number

The reference is docked five times at seed 42 before a pocket is published. If the five scores are not identical, the target is skipped and the next structure in the list is tried. For PSAP-4V2O the five runs returned -4.750 each time.

The margin

A win has to be at or below the reference minus 0.5 kcal/mol. The margin exists because a difference of 0.01 kcal/mol is inside the method's own noise and should never pay anyone. It is published per pocket and fixed when the pocket opens.