So I looked through the code and it doesn't matter if you are using the rms or dme metric, but it seems to realign to the centroid after every pass (in the means algorithm anyway) not just at the end of the algorithm once it has stopped assigning points. This means realigning after the clustering has finished doesn't help because points are assigned based on the alignment at each pass. Again someone with more knowledge than me may want to confirm this.
commented these instances out and it does what I am expecting it to do.
On Jul 23, 2010, at 10:46 AM, Daniel Roe wrote:
> Someone more knowledgeable about clustering in Ptraj may want to chime
> in here, but I suspect that you're using 'rms' as your distance
> metric, and I the RMS functionality in ptraj does indeed change around
> coordinates during the course of the RMSD calculation (that is to say
> inside the function itself). So I don't think this can be prevented.
> Maybe you could realign the frames after clustering?
>
> On Thu, Jul 22, 2010 at 9:55 PM, Geoffrey Wood <gwood.mit.edu> wrote:
>> Dear Amber Community.
>>
>> I have been doing some clustering analysis recently and was surprised to see that after ptraj performed the clustering the coordinates of the trajectory have changed. This is important to me because before the clustering was done I had aligned all the frames to 3 atomic centers ('mask1') and then performed the clustering on another part of the molecule ('mask2'). So before the clustering is done these atoms are aligned, after the clustering is done none of the clusters have these atoms aligned.
>> Is there anyway for this to be prevented?
>>
>>
>> Thanks in advance, Geoff
>>
>>
>> _______________________________________________
>> 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 Fri Jul 23 2010 - 08:30:03 PDT
