Re: vmd-l: measure fit - wrong alignment

From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Thu May 26 2011 - 13:09:29 CDT

Hi Eduardo,

The alignment in your data is optimized, the problem is the ordering
of atoms. See this message:
http://www.ks.uiuc.edu/Research/vmd/mailing_list/vmd-l/10172.html

Cheers,
Jerome

On 26 May 2011 19:17, Eduardo Oliveira <eduardoftoliveira_at_gmail.com> wrote:
> Thank you for your fast reply.
> I tried the alignment with 4, 5 and 6 atoms, all equally distant from
> each other. However, I still don't have a perfect fit, not even close.
> I attached the pdb's with 6 atoms.
>
> Thank you
>
> 2011/5/26 Jérôme Hénin <jhenin_at_ifr88.cnrs-mrs.fr>:
>> Hi Eduardo,
>>
>> In principle it is easy to align groups of 3 atoms, but some commonly
>> used algorithms only work well for 4 atoms or more. I think VMD's
>> algorithm has this limitation, the (rare) case of 3 atoms is
>> sacrificed in favor of fast and robust alignments for large
>> structures.
>>
>> Cheers,
>> Jerome
>>
>> On 26 May 2011 12:56, Eduardo Oliveira <eduardoftoliveira_at_gmail.com> wrote:
>>> Good morning
>>> I've been trying to superimpose two pdb structures using the measure
>>> fit command, but I keep having misaligned molecules.
>>> The 3 atoms I choose to superimpose are exactly at the same distance
>>> in both pdb's, so I should have a perfect alignment. I made 2 pdbs
>>> with only the 3 atoms I want to allign so you can try to replicate my
>>> problem (I want to superimpose pdb2 into pdb1).
>>> I'm using vmd 1.8.7 and I type the following commands
>>>
>>> vmd > set reference_sel [atomselect 0 "all"]
>>> vmd > set comparison_sel [atomselect 1 "all"]
>>> vmd > set move_sel [atomselect 1 all]
>>> vmd > set transformation_matrix [measure fit $comparison_sel $reference_sel]
>>> {0.6737986207008362 0.5753872394561768 -0.46359989047050476
>>> -16.962047576904297} {-0.2173646241426468 0.7539900541305542
>>> 0.6198800206184387 27.565969467163086} {0.7062209248542786
>>> -0.31690412759780884 0.6331063508987427 -12.475563049316406} {0.0 0.0
>>> 0.0 1.0}
>>> vmd > $move_sel move $transformation_matrix
>>>
>>> Thank you
>>>
>>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:24:00 CST