[Zinc-fans] Clustering and measuring diversity -reg

[John J. Irwin]

Hi Rafi

Thank you for your email and your interest in ZINC.

>  
> I have few basic questions. If we select a database for virtual
> screening, how should we analyze the properties of the database? Is
> there any parameters which should be looked into.
We do some basic property distribution analysis, which we attempt to
economically compress into four scatterplots. For example, see the "lead
like subset" graphs here...
http://zinc.docking.org/subset1/1/index.html
You may also download the properties at
http://zinc.docking.org/subset1/1/1_prop.xls
and use gnuplot (or whatever) yourself.
>  
> My other question is: How is the diversity of the databases analyzed? 
> Can anyone give me some idea about the tanimoto and how it is measured.
We have our own statistics for "diversity" which we provide. We use
SUBSET 1.0 to pick cluster representatives from a low-to-high molecular
weight sorted list. We cluster at 90, 80, 70, and 60 % Tanimoto circles
from the representatives. This gives you some idea of the amount of
"chemical similarity" present in the collection, but it is just one of
many ways to do it.

To know more about Tanimoto, please look at the Daylight Theory Manual
(google it).

>  
> Is there any free software by which we can cluster the database and
> also study the diversity. Or is there any approach to do a clustering
> analysis and take representative examples from each cluster and study
> it by further docking.
SUBSET 1.0 is free. (from Marc Nicklaus's lab). I suspect our approach
is not viewed as orthodox by the leading practitioners in the field of
similarity clustering. The thing to ask youself is: what is the
question? What do you think you are measuring or estimating with
"diversity".
>  
>  
> Sorry for asking lot of question. But your reply will be very helpful.
I am happy to be asked questions, particularly ones I have a reasonable
shot at answering. ;-)

>  
>  
> Thanks,
> Rafi
>  
>  
>  
>  
>  
>  
>  
>  
>  
>  
>  
>  
>  
> ------------------------------------------------------------------------
>
> _______________________________________________
> Zinc-fans mailing list
> Zinc-fans at docking.org
> http://blur.compbio.ucsf.edu/mailman/listinfo/zinc-fans
>