Re: AMBER: silhouette width/coefficient after clustering by PTRAJ9.9

From: Dong Xu <d1xu.ucsd.edu>
Date: Fri, 4 Jul 2008 12:21:31 -0700

Hi Jianyin,

The silhouette.pl script froze my computer with 4GB ram when I ran it on a
570MB PairwiseDistances and 12140 frames in ClusterMerging.txt. I wonder if
there's a way to estimate the memory usage and run time based on the input
filesize? Do you have any silhouette programs that are binaries compiled in
C or Fortran?

Thanks,

Dong

On Thu, Jul 3, 2008 at 11:33 PM, Dong Xu <d1xu.ucsd.edu> wrote:

> Also, when you use "Readmerge" to read ClusterMerging.txt generated by
> averagelinkage, does PairwiseDistances file have to be present? If not, will
> Readmerge re-calculate PairwiseDistances?
>
> Thanks,
>
> -DX
>
>
> On Thu, Jul 3, 2008 at 11:28 PM, Dong Xu <d1xu.ucsd.edu> wrote:
>
>> Hi Jianyin,
>>
>> I see there's an option [verbose *verb*]. Could you tell me how to use.
>> The clustering manual didn't mention it at all.
>>
>> Thanks,
>>
>> -DX
>>
>>
>> On Fri, May 30, 2008 at 10:11 AM, Jianyin Shao <jyshao2004.gmail.com>
>> wrote:
>> >
>> >
>> >>
>> >> I understand that "X" labels the correspondent frames to the cluster
>> >> members. I wrote a python script to translate "X" to frame number. But
>> >> what I wanted to know was how to obtain or extract the frame numbers
>> >> for the cluster representatives in "filename.rep.ci" files.
>> >
>> > One "dumb" trick you may try is to use the representative structure as
>> > reference and run rms fitting for the entire trajectory. The one with
>> rmsd
>> > of 0 should be the representative frame.
>> >>
>> >> Since ptraj already has "rms" command, rms fitting in clustering seems
>> >> a bit redundant if "rms" command is explicitly required prior to
>> >> running "cluster" command.
>> >
>> > The rms command will fit to a specific structure. In clustering, the
>> > structures in the trajectory will be aligned to different representative
>> > structures of clusters to obtain reasonable centroids.
>> >
>> >
>> > Good luck,
>> >
>> > Jianyin
>> >
>> >
>>
>>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo.scripps.edu
Received on Sun Jul 06 2008 - 06:07:51 PDT
Custom Search