Re: [AMBER] clustering in ptraj

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Tue, 10 Mar 2015 08:08:14 -0600

On Mon, Mar 9, 2015 at 11:39 AM, Mary Varughese <maryvj1985.gmail.com> wrote:
> on a modelled system, do you think the most populated cluster will be the
> probable one?

By definition the most populated cluster is the most probable one *for
that particular ensemble*, i.e. that doesn't mean it's the most
probable in terms of free energy. To ascertain that you need to have
some confidence that you are observing all relevant conformations with
relative populations close to their Boltzmann-weighted ones (within
the confines of whatever force field you are using), which typically
involves multiple independent simulations run to convergence. For some
examples of this kind of analysis see
http://pubs.acs.org/doi/abs/10.1021/ct400862k and
http://pubs.acs.org/doi/abs/10.1021/jp4125099.

> and would increasing the number of clusters will be better.?

Maybe - you will need to try a few different settings and look at your
cluster metrics (DBI etc). There really is no way to tell beforehand
what clustering parameters will be optimal for a given set of
structures.

Good luck,

-Dan

>
> On Mon, Mar 9, 2015 at 7:42 PM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>
>> Hi,
>>
>> Clustering tends to be more of an art form than a science, and it can
>> be difficult to tell a prior what algorithm/distance metric/clustering
>> parameters will give you optimal clusters. Usually something like
>> hierarchical average-linkage or K-means with RMS give decent results,
>> but it's not guaranteed. I highly recommend reading this article on
>> clustering MD trajectories by Shao & Cheatham III et al.:
>>
>> http://pubs.acs.org/doi/abs/10.1021/ct700119m
>>
>> -Dan
>>
>> On Mon, Mar 9, 2015 at 12:48 AM, Mary Varughese <maryvj1985.gmail.com>
>> wrote:
>> > Sir,
>> >
>> > I read that there are different options to perform clustering in ptraj.
>> > It would be helpful if you could give some suggestions on my problem.
>> >
>> > I have a nucleic acid system which is a modelled one. have run a 50 ns
>> > simulation (explicit).
>> > obviously the starting structure may be some different from the probable
>> > structure.
>> > Which clustering algorithm do you think is suitable under such situation.
>> >
>> > thanking you
>> >
>> > Mary Varughese
>> > _______________________________________________
>> > AMBER mailing list
>> > AMBER.ambermd.org
>> > http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>>
>> --
>> -------------------------
>> Daniel R. Roe, PhD
>> Department of Medicinal Chemistry
>> University of Utah
>> 30 South 2000 East, Room 307
>> Salt Lake City, UT 84112-5820
>> http://home.chpc.utah.edu/~cheatham/
>> (801) 587-9652
>> (801) 585-6208 (Fax)
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber



-- 
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 10 2015 - 07:30:02 PDT
Custom Search