[Dock-fans] similarity clustering

Ben Keshet keshet1 at umbc.edu
Wed Apr 21 05:38:39 PDT 2010


On 4/20/2010 10:11 PM, Scott Brozell wrote:
> Hi,
>
> On Mon, Apr 19, 2010 at 04:41:12PM -0400, Ben Keshet wrote:
>    
>> I have a question that is a little beyond DOCK / ZINC but some users
>> will likely know the answer:
>> I want to cluster molecules of a large library based on some simple
>> structure features. Does anyone know any academic-free software that can
>> do that?
>>
>> For example, I docked ~360,000 molecules. Some of the molecules in the
>> top 200 are actually very similar structurally, so I want to cluster
>> them into one group. This meant to allow an in-vitro testing of a
>> diverse population. I am sure that this is a pretty common task, however
>> for some reason I am having hard time finding suitable applications.
>>      
> I haven't done that, but:
> http://ligand.info/
> might be useful.
> Chimera can only do clustering of a
> conformational ensemble (where each member contains the same atoms)
> http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/ensemblecluster/ensemblecluster.html
>
> Try searching the CCL archives.
>
> Scott
>
>
>    
Thank you. I actually tried ligand-info, thought I may have missed 
something about how to use it correctly/efficiently (I have some doubts 
about the quality of results I got; there were some inconsistencies, as 
far as I could tell, depending on the library size). Since I posted the 
question here I found two additional applications I was going to try:

http://weizhong-lab.ucsd.edu/chemspace/
http://cactus.nci.nih.gov/subset/#2

Thanks for the reference to CCL, it seems like this issue was discussed 
there in different forms before.

Ben


More information about the Dock-fans mailing list