[Dock-fans] ideal parameters for screening a large compound library

Cen Gao cengao at gmail.com
Sun May 30 16:26:48 PDT 2010


Dear Dock Users:

I am new to Dock6 and I plan to use DOCK to virtual screening a large
compound library. (i.e. a subset of ZINC database) Sphere and grid were
generated using similar parameters as in the tutorial. After that, I tested
a few compounds through the flexible ligand docking protocol with the
default parameters. These compounds are randomly picked from ZINC. The
docking time (primary grid score only)  ranges from 3 mins to 30 mins per
compound, which might be a little too slow for us. (I have tested both on a
Linux cluster and a windows PC running cygwin)  The "verbose growth stats"
shows most time was spend on grown the ligands, especially the last few
layers for compounds with more than 7 rotatable bonds.

As we plan to dock a large compound library, probably we can only afford a
speed of 1 to 2 mins per compound. After reading the manual, I tried to
change the following three parameters to smaller values:
max_orientations                                             500 --> 400
pruning_max_orients                                        100 --> 50
pruning_clustering_cutoff                                  100 --> 50


In this case docking does become faster, but still cost us more than 5~10
mins for most compounds. I can change the value to even smaller but I am not
so sure how far I can go for this without sacrificing too much on accuracy .
So my question is:

1. With the default parameters for flexible docking, is my time (3 mins ~ 30
mins per compound ) normal for DOCK6?

2. Are there any recommended parameters for screening a large compound
library? Or if anyone has done this before, would you mind to share your
parameters?


Thank you


Cen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.docking.org/pipermail/dock-fans/attachments/20100530/1225e49b/attachment.html 


More information about the Dock-fans mailing list