From: Wong Li Zhe (Wong.LiZhe_at_student.imu.edu.my)
Date: Thu Dec 08 2016 - 18:37:12 CST

Thank you for the reply, Ashar and John!

Just one quick question, would it be possible to differentiate the superimposed structures with colour code? (From the Superpose output)

For instance, structure A in blue and structure B in red?

Best regards,

Li Zhe

________________________________
From: Ashar Malik <asharjm_at_gmail.com>
Sent: Wednesday, 7 December, 2016 12:56:55 PM
To: Wong Li Zhe; vmd-l_at_ks.uiuc.edu; John Stone
Subject: Re: vmd-l: Superimpose

Sorry accidentally replied to just you instead of the list.
The thread is intact ...

Following on your latest set of questions:

That is not the superpose I was talking about.
Use it from here if you don't want to install :
http://www.ebi.ac.uk/msd-srv/ssm/
Hit the launch PDBeFOLD button. On the following page. Select coordinate files from the dropdown and upload both pdb files that you want to compare.
In the results summary page - if you click the index you will see a rotation matrix like this

 -0.421 -0.104 -0.901 X -16.794

  0.199 -0.980 0.021 Y + 26.754

 -0.885 -0.170 0.433 Z 37.528

Remove the 4th column -- and set up a matrix to reflect the values from your alignment.

Well, when I input the 2 protein structures with different number of residues, PROMALD generated 2 PDB files based on their alignment.

What I have done is that, I then open the 2 PDB files and look for matching residues sequence. For instance, residue ILE, ARG, ASN, LEU are in position 1000, 1001, 1002, 1003 of protein A and 534, 535, 536, 537 of protein B.

Since you persist on doing it the hard way :: follow my last email.

Once you have selected that 1000 should line up with 534 and 1001 with 535 etc

start VMD
load the two structures (first load the structure which has residues 1000,1001 etc and then the other)
since you know the residues across the two structures are identical now you can remove the backbone keyword because identical residues mean identical number of atoms.
use the following

set prot_0 [atomselect 0 "resid 1000 1001 1002 1003 ..."]

put all your residues numbers in the above selection from 1 of the structures

set prot_1 [atomselect 1 "resid 534 535 536 537 ... "]

put all your residues numbers in the above selection from the other of the 2 structures

set f [measure fit $prot_0 $prot_1]
set r [measure rmsd $prot_0 $prot_1]
puts "RMSD between selections is :::"$r

The above will return an RMSD value between your selections

set all_0 [atomselect 0 all]
$all_0 move $f

the last two commands will put one structure on top of the other which is what you want.

Those protein residues sequence that doesn't match, I deleted them based on the alignment file generated.

Please don't delete lines/residues from the PDB file. When you will load it back into VMD - there will be GAPS and I don't see the point of viewing any alignment like that.

On Wed, Dec 7, 2016 at 5:55 PM, Ashar Malik <asharjm_at_gmail.com<mailto:asharjm_at_gmail.com>> wrote:
Sorry accidentally replied to just you instead of the list.
The thread is intact ...

Following on your latest set of questions:

That is not the superpose I was talking about.
Use it from here if you don't want to install :
http://www.ebi.ac.uk/msd-srv/ssm/
Hit the launch PDBeFOLD button. On the following page. Select coordinate files from the dropdown and upload both pdb files that you want to compare.
In the results summary page - if you click the index you will see a rotation matrix like this

 -0.421 -0.104 -0.901 X -16.794

  0.199 -0.980 0.021 Y + 26.754

 -0.885 -0.170 0.433 Z 37.528

Remove the 4th column -- and set up a matrix to reflect the values from your alignment.

Well, when I input the 2 protein structures with different number of residues, PROMALD generated 2 PDB files based on their alignment.

What I have done is that, I then open the 2 PDB files and look for matching residues sequence. For instance, residue ILE, ARG, ASN, LEU are in position 1000, 1001, 1002, 1003 of protein A and 534, 535, 536, 537 of protein B.

Since you persist on doing it the hard way :: follow my last email.

Once you have selected that 1000 should line up with 534 and 1001 with 535 etc

start VMD
load the two structures (first load the structure which has residues 1000,1001 etc and then the other)
since you know the residues across the two structures are identical now you can remove the backbone keyword because identical residues mean identical number of atoms.
use the following

set prot_0 [atomselect 0 "resid 1000 1001 1002 1003 ..."]

put all your residues numbers in the above selection from 1 of the structures

set prot_1 [atomselect 1 "resid 534 535 536 537 ... "]

put all your residues numbers in the above selection from the other of the 2 structures

set f [measure fit $prot_0 $prot_1]
set r [measure rmsd $prot_0 $prot_1]
puts "RMSD between selections is :::"$r

The above will return an RMSD value between your selections

set all_0 [atomselect 0 all]
$all_0 move $f

the last two commands will put one structure on top of the other which is what you want.

Those protein residues sequence that doesn't match, I deleted them based on the alignment file generated.

Please don't delete lines/residues from the PDB file. When you will load it back into VMD - there will be GAPS and I don't see the point of viewing any alignment like that.

________________________________
From: Ashar Malik <asharjm_at_gmail.com<mailto:asharjm_at_gmail.com>>
Sent: Wednesday, 7 December, 2016 10:32:14 AM
To: Wong Li Zhe
Subject: Re: vmd-l: Superimpose

Hi,

Not sure why but I didn't get any transformation matrix from Superpose, hence, I couldn't view it using VMD.

Did you use superpose from the CCP4 suite of programs here: http://www.ccp4.ac.uk/html/superpose.html
Once it is installed properly you type

superpose protein1.pdb protein2.pdb

A lot of stuff will flash on the screen. In the stdout you will find a few lines like

Query protein1.pdb
and Target protein2.pdb

 have been superposed. Superposition matrix (to be applied
 to ref.pdb) is

        Rx Ry Rz T
      1.000 0.000 -0.000 -0.000
      0.000 1.000 -0.000 0.000
     -0.000 -0.000 1.000 -0.000

 If you open protein1 and protein2 in VMD and then apply the above matrix from superpose onto protein1 it should superimpose the two proteins in VMD graphics window for you.

Anyway, I used another server (PROSMALS3D) to run alignment and etc as what has been suggested by you.

I don't remember suggesting PROSMAL!!!!

Did you get an output like in the attached PDF.

Managed to edit the PDB file manually using Excel and removed those unwanted residues.

What do you mean removed the unwanted residues? Did you just delete the lines? If you were just deleting lines - you could have just deleted them in the PDB file (open the PDB file in a text editor - like gedit/nano). Which lines are you deleting ? Are you deleting those which are not aligning sequentially, because the output that I have attached is a sequence output. This server that you are using produces a sequence alignment --- even if you do cut down by deleting lines/residues - the remaining residues will NOT be equal as they STILL have different number of atoms. VMD will not superpose unless you have an equal number of atoms. You should refer to my previous email and do selections manually on the backbone atoms only. If you don't to and want to continue using your method you can delete lines but for the structure in VMD you would still have to select backbone atoms - so that the two selections are equal. I don't understand why you would want to do this. Because the deleted lines means an incomplete structure will be aligned - I don't see how that will offer any useful information. You could use the transformation matrix from aligning the edited structure on the whole. But that would give you the same answer which I have already told you ....

from both structures select equal number of residues and from them select backbone atoms only. superimpose them.

Prior to superimpose both protein structures on VMD, I tried converting from Excel to PDB but somehow it doesn't work with all those funny alphabets present in the file.

A couple of things here. 1) A PDB file is a highly formatted file which each column assigned to a certain value. see here: http://deposit.rcsb.org/adit/docs/pdb_atom_format.html
Look at the Record : Atom entry. There are more details in that and an example read it. 2) How did you convert the excel file to PDB ? PDB is just a text file. Did you save it as an excel file and open that excel file with a word editor? That would obviously not work because Excel has its own format - which is why you may see unfamiliar characters.

I wonder do you have any insights on this?

If you want to continue using PROMALS3D - you can get the residues that align -- so e.g. if your file tells you that residues 1,20,22,34,56 from protein 1 align with 3,21,22,35,57 in protein 2 -- (notice they are equal in number of residues - both 5 in number) --

start VMD
you should load protein1 into VMD
you should load protein2 into VMD
set prot_0 [atomselect 0 "resid 1 20 22 34 56 and backbone"]
set prot_1 [atomselect 1 "resid 3 21 22 35 57 and backbone"]
set f [measure fit $prot_0 $prot_1]
set r [measure rmsd $prot_0 $prot_1]
puts "RMSD between selections is :::"$r
set all_0 [atomselect 0 all]
$all_0 move $f

like i said in my previous email - YOU DO NOT NEED to delete lines/edit/use excel/convert or anything of that sort.
Use promals3d to find which residues you wait to align. once you have those - use the code I gave you - change the number of residues to match what you get from promals3d. use the code. You will get your answer.

I have tried online conversion and it doesn't work.

Any thoughts on this?

Thank you in advance.

Best regards,

Li Zhe

--
Best,
/A
--
Best,
/A
--
Best,
/A