AMBER Archive (2008)

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

From: Dong Xu (
Date: Tue May 27 2008 - 13:20:09 CDT

On Tue, May 27, 2008 at 9:03 AM, Jianyin Shao <> wrote:
> On 5/27/08, Jianyin Shao <> wrote:
>> On 5/25/08, Dong Xu <> wrote:
>>> Hi Jianyin Shao,
>>> The script works great!
>>> A question about decoy:
>>> First I run averagelinkage to determine the cluster count and to
>>> obtain the cluster representatives. Then I use the cluster count and
>>> the cluster representatives from averagelinkage as the seed structures
>>> for decoy. You mention that decoy_structure file should be a
>>> trajectory file. Does it have to be in amber mdcrd format? What's the
>>> best way to convert multiple pdb file to an amber (or maybe other)
>>> trajectory file that is accepted by decoy?
>> You can use another ptraj script:
>> trajin pdb1.pdb
>> trajin pdb2.pdb
>> trajin pdb3.pdb
>> ...
>> trajout decoy.ptraj pdb
> Oops. Should not specify pdb when outputting trajectory file.
> Use "trajout decoy.ptraj" instead.
> Sorry for the mistake.

If I want to fix the seed structure that I provide for decoy, I will
have to specify "iteration 1", correct?

So I tried it and compared my input seeds (averagelinkage cluster
representatives) and decoy output representatives. They look
different. I remember the paper mentioned that the representative are
not the centroid, but a structure close to the centroid instead. Is
that the reason why representatives deviate from the seeds even if I
use "iteration 1"? Centroid is not necessarily an actual frame from
the trajectory, but a representative is, correct?

Sorry for asking so many question. I'm trying to use your ptraj
clustering facility on my upcoming paper and I think it is a really
powerful tool! Thanks so much for answering my question!

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)