[Biopython] superimposing problem

[João Rodrigues]

The problem here is that for that picture to be possible you have to have
exactly similar flanking residues. And in those structure of yours, it seems
you don't.

Peter was trying to find a way of making sure that the superimposition was
unambiguous. In other words, that the loops superimposed in the only way
possible (rotation- and translation-wise). By having more than one point per
residue, as you are doing with those CA, C, N and O atoms, you're getting
the best superimposition possible. However, since the ends of the loop do
not match, you can never fit them as in the picture.

I'm assuming you're generating these loops out of loop modelling program of
sampling program. Try adding dummy residues before and after the loops, say
two ALAs, and fix them during the procedure. If, on the other hand, you want
a good measure of displaying loop flexibility/variability, try superimposing
only the first residue. That should be unique enough to display the
different conformations.

Sorry I can't be of much help...

João [...] Rodrigues
http://doeidoei.wordpress.com



On Wed, Dec 8, 2010 at 12:18 AM, George Devaniranjan <devaniranjan at gmail.com
> wrote:

> Hello Peter,
> Thank you for your suggestion.
> I am using CA/C/N/O  for superimposition of S1/S2 and E1/E2.
>
> Not sure I got exactly what you meant but I am going to think about it
> tonight  and see if that makes sense.
>
> The idea that I am trying to see is ---if 2 coils from diff protein have
> approximately the same length (S1-E1  = S2-E2 distance) and have the same
> number of residues what is the "floppiness" or (no of conformations) that
> you can see.
>
> So as in my figure I sent (reattached) it does not matter where the
> intermediate residues are they can flop about.
> Thanks I will think about this more in the next few days.
> Regards,
> George
>
>
>
>
> On Tue, Dec 7, 2010 at 6:00 PM, Peter <biopython at maubp.freeserve.co.uk>wrote:
>
>> On Tue, Dec 7, 2010 at 10:49 PM, George Devaniranjan
>> <devaniranjan at gmail.com> wrote:
>> > Thanks João for trying.
>> > I was also thinking that you would have to deform the 2nd coil to get
>> > exactly what I want.
>> >
>> > What I tried and not completly sussefully but  I will try that agian is
>> as
>> > follows.
>> >
>> > 1) Superimpose 1st residue of coil1 with 1st residue of coil2, rot/tran
>> > coil2 and get the coordinates---lets call the rot/tran coil2   as
>> mod_coil2
>> >
>> > 2) Superimpose last residue of coil1 with last residue of mod_coil2 and
>> then
>> > rot/tran whole molecule---the result is not exactly what I want but it
>> is
>> > the only solution I can think of for now.
>> > Thanks once agin João,
>> >
>> > George
>>
>> Hi George,
>>
>> Hopefully I have understood your aim, and the following makes
>> sense...
>>
>> Have you thought about (in your head - no code needed) trying to
>> simultaneously try to superimpose the first residues AND the last
>> residues with a rigid body motion (rotation and translation)?
>>
>> Assuming you are using just the C-alpha atoms for this. Lets
>> call these atoms S1, E1 and S2, E2 (start and end). Also
>> suppose that the distance S1 to E1 is bigger than S2 to E2.
>> The superposition will  give you S1, S2, E2, E1 on a line in space.
>> However, the relative rotation of the two loops is free.
>>
>> So, if using C-alpha atoms,  I think you are going to have to include
>> at least one pair of atoms from the loop, e.g. M1 and M2 for the
>> C-alpha of the mid point residue.
>>
>> However, if you are using more than just the C-alpha atoms, that
>> should give enough constraints for the superposition to be unique.
>> It may still not be what you want, in which case try adding more
>> constraints as above.
>>
>> Peter
>>
>
>