How to superimpose two proteins with SPDBV.

In this example, we will quickly describe the manual method used to superimpose proteins that share the same fold when the automatic superimposition method fails. Commands to apply on Swiss-PdbViewer appear in bold.

 

Open the PDB file 11MUP

Open the PDB file 21OBP

Color by secondary structure (hold the shift key when invoking the command to apply the command on both proteins.

As you can see, those two structure are quite similar, and share a jelly-roll fold. Assuming you want to compare those two proteins, the first step is to superimpose them in 3D-space. For proteins sharing a high similarity, the "Iterative Magic Fit" option of the "Tools" menu is the best approach
The primary structures of the two proteins are aligned using SIM. The best matching fragments (similar amino-acids) are used as a seed to a 3D match algorithm. However, when too many amino-acids that do not really correspond in 3D space have been paired by SIM, the method occasionally fails. This will be the case for this example.

Use the "Iterative Magic Fit" option of the "Tools" menu

As you can see, 21OBP has moved onto 11MUP, but the superimposition is completely wrong (which is signalled by the fact that SPDBV is unable to derive a structural alignment):

Why?

Here is the alignment proposed by SIM. The three matching fragments are indicated with [---] and are coloured in red green and blue respectively. Those fragments have been used as "seed" to the 3D matching algorithm.

 

                           [--------- ---------- ---------- ----------
11MUP       1   EEASSTGRNF NVEKINGEWH TIILASDKRE KIEDNGNFRL FLEQIHVLEN
21OBP.aa    1       EEEAEQ NLSELSGPWR TVYIGSTNPE KIQENGPFRT YFRELVFDDE
                        .  *.  ..* *. *. ..* . * **..** **  .. ..   ..
 
                ------] [- ---------- ---------- ---------- ----------
11MUP      51   SLVLKFH-TV RDEECSELSM VADKTEKAGE YSVTYDGFNT FTIPKTDYDN
21OBP.aa   47   KGTVDFYFSV KRDGKWKNVH VKATKQDDGT YVADYEGQNV FKIVSLSRTH
                   . *  .* . .        *    .  *  *   *.* *  * *      .
 
                ---------- ---------  ---] [---- ---------- ----------
11MUP     100   FLMAHLINEK DGETFQLMGL YGREPDLSSD IKERFAQLCE EHGILRENII
21OBP.aa   97   LVAHNINVDK HGQTTELTGL FVKL-NVEDE DLEKFWKLTE DKGIDKKNVV
                ..  ..  .*  *.* .* ** . .  ..  .   *.*  * * . ** . *..
 
                -]
11MUP     150   DLSNANRC                                              
21OBP.aa  146   NFLENEDHPH                                            
                .. . .

 

Unfortunately, only the first matching fragment (the red one) is meaningful. The next version of SPDBV will be able to detect automatically the best fragment to use in those situations, but right now we need to manually select matching residues in order to do a "preliminary 3D-match" that can subsequently be improved.

Show Align window.

Amino-acids that were selected as "seed" to the 3D match appear in inverse video: they are selected. As mentioned above, too many amino-acids were selected (almost all residues were selected). As a first approach, we will select residues 11-27 in 11MUP and residues 7-23 in 21OBP. They include a small helix and a strand, and "look similar". Basing the selection on secondary structure is usually sound. Note that it is very important to select the same number of residues in each sequence, as each amino-acid must have a correspondent in the other sequence:

Now use the "Fit Molecules (auto) item of the "Tool" menu.

The rms fit for those 17 Ca is 1.52Å As you can see, the two proteins are now quite well superimposed in 3D space. However, this superimposition can be improved. Indeed, more than 17 residues are similar in the two sequences, and we want to select as many residues as possible before doing the superimposition. This can be done automatically:

use the "improve fit" item of the "Tools" menu.

The rms fit for those 17 Ca is 1.52Å As you can see, the two proteins are now quite well superimposed in 3D space. However, this superimposition can be improved. Indeed, more than 17 residues are similar in the two sequences, and we want to select as many residues as possible before doing the superimposition. This can be done automatically:

use the "improve fit" item of the "Tools" menu.

Now the proteins are really well superimposed (at least the b-barrel). Look at the structural alignment, and note that less residues are selected than with the automatic Magic fit option. C-terminal residues are not selected, as they are very far apart in 3-D space. Hence they were not used during the 3D-match, which explains why this time the match is successful.

Now select layer 21OBP (click on the word 21OBP in the Align window) and color the protein by its RMS deviation to 11MUP (color menu). At this point, each amino-acid of the active protein will be colored accordingly to its RMS backbone deviation from the corresponding amino-acid of the reference protein (the first loaded). Dark blue means good superposition whereas red means bad superposition.

Now toggle the displayed molecule with Control tab. This with alternately display only one protein, and is very convenient to detect where the proteins are the most similar. Remark that the central core of the protein (the jelly-roll) is blue (low RMS deviation), whereas external loops appear in red, meaning there is no counterpart in 11MUP.

 

[ Back to Index | Next Chapter ]

Last modification: Please note that these pages have not been updated since 1999 and are provided on a "as is basis". We are currently working on an updated version of this course.