Hi Senthil,
I am not surprised that you see RMSD > 1 in a cluster especially for all
atoms for 300 residues. If you are looking for low RMSD variation within a
cluster, you can run clustering command by using "epsilon 1.0" instead of
"clusters 5", but you may end up with more clusters. Y. The clustering will
stop merging if the distance between the two nodes that will be merged next
is larger than 1.0. And even that would not guarantee that all frames in the
cluster are within 1.0 of the representative frame.
Hope it helps.
Jianyin
On Wed, Sep 1, 2010 at 5:01 PM, Senthil Natesan <sen.natesan.yahoo.com>wrote:
> Hi Jianyin,
>
> Thanks for the detailed and prompt reply.
> In my analysis, I found that there are 5 clusters and the last one has more
> number of points.
> I took the representative pdb from that cluster and used as a reference
> and
> calculated RMSD for
> all the other points in the cluster. I was expecting very low RMSD, but
> almost
> all points
> have RMSD > 1 to this representative structure.
>
> Can you guide me how to interpret this result?.. It would be great if you
> can
> suggest me
> better methods,(if any) to efficiently use this program.
>
> I would be very thankful if anyone can point me best available methods to
> analyze trajectories.
> My ultimate goal is to get a representative structure that is very close to
> time-average structure ( I am aware
> that time-average structure is kind of meaningless in terms of coordinates)
>
> thanks,
>
> Senthil
>
>
>
> Hi Senthil,
>
> Unfortunately, you cannot visualize the ClusterMerging.txt by xmgrace. Each
> line of this file records one step of merging. The first number should be a
> negative number, which represents the newly formed node. The next two
> numbers are the nodes to be merged, if a number is positive, that is the
> frame number in the trajectory. The fourth number is distance between the
> two nodes. The next two numbers are the DBI and pSF values for current
> clustering. The DBI value will not be calculated if cluster number is
> larger
> than 50. If you want to plot the cluster # vs. DBI or pSF, you can get the
> values from the ClusterMerging.txt file without run multiple clustering
> processes.
>
> If you use sieve, the cluster merging is executed only in the first scan,
> i.e. 1 frame out of every 10 frames in your case. The rest of frames will
> be
> assigned to the closest cluster out of the 5 clusters formed in the first
> scan. Thus, information in the ClusterMerging.txt would be irrelevant to
> the
> entire trajectory.
>
> Hope this helps,
>
> Jianyin Shao
>
> On Wed, Sep 1, 2010 at 3:14 PM, Senthil Natesan <sen.natesan.yahoo.com
> >wrote:
>
> >
> > Hi Amber Users,
> > Greetings.
> > Can someone please let me know how to visualize and infer the output file
> > (ClusterMerging.txt)
> > from ptraj-cluster program ?
> >
> > I just fired up the file with xmgrace and I don't get much.
> >
> > FYI:
> > I have run a MD simulation of protein-ligand complex and used cluster
> > program from ptraj for getting representative and average structure of
> the
> > MD
> > run.
> > I used averagelinkage method for clustering. Clustering process went
> well.
> > I
> > used the
> > following command in ptraj
> >
> > cluster out mk2mdrun all amber representative pdb average pdb
> > averagelinkage
> > sieve 10 clusters 5 rms mass :1-300
> >
> > Thanks in advance.
> >
> > Senthil Natesan
> >
> >
> >
> > _______________________________________________
> > 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 Sep 02 2010 - 16:30:04 PDT
