The command line used in the tutorial lead to all searches ending while they where still actively optimizing their solutions. Therefore the first natural step is to increase the number of evaluations to allow each search to converge. Using the default value of 2.5M evals is a good idea because, unlike AutoDock4, ADFR will not only use up as long as improvements are seen and can be hoped for.

Additionally, providing the crystal structure as a reference (-r/–reference option) will results in the RMSD between the docked pose and crystal structure to be reported.

Finally, we name the job rigid1 to avoid adfr complaining that output files with the same name already exist and stopping the calculation. The -O option could be used to tell adfr to overwrite existing files silently.

The run takes longer now but all runs converge as seen below

Furthermore, all runs converge to the same solution

The ref. rmsd column shows that the docked pose is within 0.3 Angstroms of the crystal structure, indicating that for this simple ligand, 8 independent searches are sufficient to reliably identify the crystal structure pose.

The default values of 50 searches is reasonable for docking most small molecules.

We can inspect the content of the .dro file to find out how many evaluations have been used up by each search.

First unpack the file

Next, extract the ending summary line from the log file of each search.

The nbEvals column shows that searches use up different number of evals, ranging from 258’364 to 1’374’668 with the run leading to the best solution (run 004) using up 861’056 evals.