Cross docking test with rDock
Last time i tested the basics of the docking program rDock; Installation, basic setup and docking, as well as had a brief walkthrough about how to post-process the results with the rDock SD file utilities. However, as I’ve covered before, you should never use re-docking as an estimation of docking power, so this time I’ll test rDock further with a cross docking experiment using my pet PDB receptor pair 1OYT and 1G32. For an example of the installation and usage, please refer to my previous blog post.
This time I dock the ligand from PDB file 1OYT into the receptor from PDB file 1G32. Autodock Vina and Smina fails with the default docking and scoring function, but by tweaking the function with machine learning it can succeed. So how will rDock fare in this more rigoureous test?? Read on below….
The 1g32 receptor file was this time obtained from the Astex non native set, while the ligand was the same as before. The active site file was generated and the docking was performed as before, with the only difference being the receptor .mol2 file.
rbcavity -r 1OYT-1G32-crossdock.prm -was > 1g32_cavity.log rbdock -r 1OYT-1G32-crossdock.prm -p dock.prm -n 100 -i 1oyt_ligand.sdf -o 1oyt_1g32_crossdock > 1oyt_1g32_crossdock.log sdsort -n -f'SCORE' 1oyt_1g32_crossdock.sd > 1oyt_1g32_crossdock_sorted.sd sdrmsd 1oyt_ligand.sdf 1oyt_1g32_crossdock_sorted.sd -o oyt_1g32_crossdock_rmsd.sd > rmsds.log sdreport -cRMSD,SCORE 1oyt_1g32_crossdock_rmsd.sd > scores.csv pymol 1oyt_1g32_crossdock_rmsd.sd 1g32.mol2 1oyt_ligand.sdf
The docking results from the cross docking to 1G32 are not as clean as the results obtained with re docking to 1OYT. However, the best scoring pose is still only 0.61 Å off in RMSD from the native pose. The scatter plot with the correlation between score and RMSD still shows a clear discrimination between the cluster of correct poses and the poses with an RMSD > 2 Å:
This is promising as Autodock Vina had trouble with this particular cross docking, but more extensive docking experiments towards the whole astex non native set should be performed before a final judgment can be made. There is a possibility that rDock fares well with this particular receptor-ligand combination, whereas Vina/Smina/Vinardo is better with other receptors.