ADL: dealing with lots of rotatable bonds ?

Jeffrey Dyason j.dyason at griffith.edu.au
Sun Jan 15 20:13:42 PST 2006


Hi,
        I was wondering if I could get some advice on dealing with 
molecules with a large number of rotatable bonds. i.e. around the max of 
32 for Autodock. I have successfully docked small molecules (only 4 or 5 
rotatable bonds using the default autodock (LGA) parameters, except 
boosting ga_num_evals to 1,250,000). I am now trying to autodock tetra and 
penta saccharides and the results seem to suggest that the conformational 
space of the saccharide is not being adequately searched. i.e. it appears 
to stay close to a linear shape with no bending to fit the binding site. 
>From an article on blind autodocking of peptides (Hetenyi, C.; van der 
Spoel, D. Efficient docking of peptides to proteins without prior 
knowledge of the binding site. PROTEIN SCIENCE 2002, 11, 1729-1737.) I 
tried increasing the ga_pop_size to 250 and the ga_num_evals to 
10,000,000. They also suggested a ga_run of 100, I have only tried 10. 
This seemed much better than the default parameters but the docked 
conformations are still not optimal.
So are there any other parameters I should change  to help access as much 
of the conformational space as possible ? or should I try to increase the 
parameters (ga_pop_size & ga_num_evals) I have already tried ? It is 
already taking 24 hours to generate 10 conformations of my MacOSX G5, I 
will try upping this to 100 as per the above paper but I won't see those 
results for another 10 days or so. Any hints/suggestions appreciated.

Thanks Jeff.
-----------
Jeff Dyason                                    e-mail: 
J.Dyason at griffith.edu.au
Institute for Glycomics http://www.griffith.edu.au/glycomics
Griffith University                          (Gold Coast Campus)  
PMB 50 Gold Coast Mail Centre       Tel: +61 7 5552 7028
Queensland 9726 Australia              Fax: +61 7 5552 8098



More information about the autodock mailing list