Here's the thing about clustering - it's very much an art form, so
input that works for one system may not work well for another.
You should at minimum choose the algorithm, (for hierarchical
agglomerative it's 'hieragglo [epsilon <e>] [clusters <n>]'), distance
metric, and some output options. The 'clusters' and/or 'epsilon' are
the parameters that will most affect your clustering results, so be
prepared to do some experimenting. You'll want to pay attention to
your clustering metrics in your info file (DBI, pseudo-F) and adjust
your input parameters to improve them. If you haven't already, see
"the clustering paper" from Shao & Cheatham et al. for a more in-depth
discussion: https://pubs.acs.org/doi/10.1021/ct700119m
Clustering is one of those things that I really want to devote time to
creating a good tutorial one day - if I can find the time...
Hope this helps, even if it's just a little.
-Dan
On Mon, May 6, 2019 at 4:28 PM Debarati DasGupta
<debarati_dasgupta.hotmail.com> wrote:
>
> Hi Daniel,
>
> Could you send me a sample input file (which can do hierarchical agglomerative based clustering)
>
> I am finding it hard to implement it using the manual of Amber18
>
> Thank you in advance!
>
> Regards
>
>
>
>
>
>
>
>
>
>
> From: Daniel Roe<mailto:daniel.r.roe.gmail.com>
> Sent: Tuesday, April 30, 2019 9:43 AM
> To: AMBER Mailing List<mailto:amber.ambermd.org>
> Subject: Re: [AMBER] Any help on how to implement it in cpptraj?
>
> Hi,
>
> The 'leader' algorithm is not implemented in cpptraj. From my limited
> knowledge of it, probably hierarchical agglomerative is the closest
> one in cpptraj (although you may want to try others, like k-means).
>
> The distance-RMSD metric is available in cpptraj, although in practice
> I have not found it much different from basic RMSD with fitting.
>
> Hope this helps,
>
> -Dan
>
> On Fri, Apr 26, 2019 at 12:25 PM Debarati DasGupta
> <debarati_dasgupta.hotmail.com> wrote:
> >
> > I am trying to do ---->Analysis of MD simulations and clustering procedure.
> > I need to implement the leader algorithm for clustering according to the distance root mean square between two MD snapshots a and b,
> > which was calculated using the intermolecular distances dij between pairs of non-hydrogen atoms in acetonitrile and eight residues in the ABL-KINASE active site. A DRMS threshold of 1 ˚A is needed for clustering by the leader algorithm. The DRMS calculation does not require structural overlap."
> >
> > Any input files how to implement it in cpptraj AMBER18?
> >
> > I am trying to reproduce the methodology of " Small Molecule Binding to Proteins: Affinity and Binding/Unbinding Dynamics from Atomistic Simulations Danzhi Huang* and Amedeo Caflisch*"
> >
> > Debarati
> >
> > _______________________________________________
> > 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
>
> _______________________________________________
> 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
Received on Thu May 09 2019 - 12:00:03 PDT
