April 2007 - Chimera-users - RBVI Mailing Lists

Re: [Chimera-users] [chimera-dev] Building a multimer model
by Elaine Meng 07 Apr '07

07 Apr '07

Hi Benjamin, There is probably more than one way to do this in Chimera, but I'd open the protein of interest six times and match it to the six monomers of the related protein. Let's call your protein "A" and the similar protein "B" (for which a hexamer is available or can be easily constructed from matrices in the PDB file). Since you didn't mention the specifics, I didn't try the whole process outlined below (not having a good example to work with), but have used the tools in other contexts. This is kind of a brute force approach. Someone else may know of a more elegant way... Making the hexamer of B: If the PDB file doesn't already have a hexamer, it should have matrix information that allows it to be constructed. If BIOMT, use Multiscale Models (under Tools... Higher-Order Structure). Initially, this doesn't copy all the atoms but just makes surfaces for computational expediency. Since you will need copies of all the atoms for matching, using the Multiscale Models GUI: select "All" chains, change their style to anything other than surface (e.g. Show... Ribbons). You can see the model IDs of the copies of B in the Model Panel (under Favorites). If SMTRY, CRYST1, or MTRIX you could use the Unit Cell tool (under Tools... Higher-Order Structure) instead of Multiscale Models. Relevant manual pages: http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/multiscale/ framemulti.html http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/unitcell/ unitcell.html Matching the 6 copies of A to the monomers of B: Options for matching are the "match" command or MatchMaker (GUI under Tools... Structure Comparison, also implemented as a command, "matchmaker"). If you wanted to name specific atoms to use for fitting, you'd use the "match" command. That can be tedious, and I'd try MatchMaker first - if there is some sequence similarity, even fairly low, it usually works quite well. By default, it iterates the fit so that only residues that match well in space are ultimately used to superimpose the proteins. That means the parts that are different should not interfere with a good fit of the parts that are the same. You can adjust several parameters, but for example if B is model 0, chains A-F and copies of A are models 1-6, these commands would use the matchmaker defaults: matchmaker #0:.a #1 pair ss matchmaker #0:.b #2 pair ss matchmaker #0:.c #3 pair ss matchmaker #0:.d #4 pair ss matchmaker #0:.e #5 pair ss matchmaker #0:.f #6 pair ss (pair ss just means to use the specified chains rather than trying all chain combinations between the models) You could specify other parameters with various keywords, for example the stringency to use when iterating the fit, how much weight secondary structure should have vs. the residue types in the sequence, etc. Relevant manual pages: http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/matchmaker/ matchmaker.html http://www.cgl.ucsf.edu/chimera/docs/UsersGuide/midas/matchmaker.html http://www.cgl.ucsf.edu/chimera/docs/UsersGuide/midas/match.html I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. meng(a)cgl.ucsf.edu UCSF Computer Graphics Lab and Babbitt Lab Department of Pharmaceutical Chemistry University of California, San Francisco http://www.cgl.ucsf.edu/home/meng/index.html On Apr 5, 2007, at 12:10 AM, Benjamin G.S. Horsman wrote: > Hello, > > My name is Benjamin Horsman and I am a student at the University of > Toronto. I have what should be a basic (possibly naive) question > regarding protein modelling that I was hoping you might be able to > answer for me or point my in the right direction. I have been > studying the structure of a bacterial protein complex, trying to > understand how it interacts with several known binding partners. The > complex is known to be a hexameric homomultimer. Recently the > structure of the subunit protein was solved. However, only the > monomeric form of the protein was determined, and the corresponding > PDB file does not contain coordinates for the multimer (i.e., no > BIOMT1, BIOMT2, BIOMT3 matrix values). I would like to assemble a > model of the hexameric complex for visual inspection and for some > docking studies. What would be the best way to accomplish this? > > There are several structures in the PDB of hexameric protein complexes > composed of subunits with very close structural alignment to my > protein of interest, differing in two regions of the protein that are > outside of the multimer binding regions. I believe that my protein > complex posses very similar or nearly identical geometry to these > known complexes. Is there any way to basically insert my protein into > this geometry? Can this be done in Chimera? If not, is there any > program or web server that allows this to be done? Or are you aware > of a better way to get this done? > > Thank you for your time. Any help is greatly appreciated. > > > Benjamin Horsman > _______________________________________________ > Chimera-dev mailing list > Chimera-dev(a)cgl.ucsf.edu > http://www.cgl.ucsf.edu/mailman/listinfo/chimera-dev

2 1

Re: [Chimera-users] viewing & comparison of multiple overlay structures
by Roberta King 04 Apr '07

04 Apr '07

Thank you Eric for your very detailed answers to my several questions. It will take me a few days to work through the suggestions. If I have any more troubles, I will contact you next week. Thanks very much, Roberta King ******************************** Roberta S. King, Ph.D. Department of Biomedical and Pharmaceutical Sciences College of Pharmacy University of Rhode Island http://www.uri.edu/pharmacy/faculty/king _____ From: Eric Pettersen [mailto:pett@cgl.ucsf.edu] Sent: Tuesday, April 03, 2007 7:57 PM To: Roberta King Cc: Chimera BB Subject: Re: [Chimera-users] viewing & comparison of multiple overlay structures On Apr 3, 2007, at 1:47 PM, Roberta King wrote: First, After overlaying multiple protein (pdb) structures with ligands using MultiAlign based on alignment, how does one easily view only the overlaid ligands? [Note: I have up to 36 structures overlaid, each with 0-2 ligands.] Hi Roberta, Using the command line (Favorites->Command Line), typing "~disp; show ligand" will hide all structures and then show only the ligands. If every structure had a ligand you would only need "show ligand" since the show command hides parts of structures that aren't to be shown, but if no part of a structure matches the show criteria, the structure isn't changed -- so you need the "~disp" also to guarantee that ligand-less structures get hidden. This is also doable with the menus by Select->Structure->ligand followed by Select->Invert Selection (or shift-right-arrow instead to invert the selection in all models) and then Actions->Atoms/Bonds->hide. From there, you will probably want to select the ligand clump by control dragging the mouse across the clump (there will be a green box outline as you drag the mouse) and then use Actions->Focus to zoom in on the clump and make it the center of rotation. Secondly, can chimera then calculate a solvent-accessible closed surface for the protein cavity surrounding each ligand? Alternatively, I could calculate the cavity first, then overlay based on alignment. The goal is to view the overlaid cavities for visual comparison. It depends. If the cavity is actually completely enclosed then Chimera won't depict it. If there is some channel to the binding pocket, then it should be depicted. I say "should" because the surfacing library we are currently using (MSMS) is known to have problems when there are narrow channels in a surface. We are working on a replacement library that should solve both these problems, but it's not ready yet. Providing that MSMS doesn't fail, your pockets should be depicted. It is also sometimes possible to work around the MSMS problem by adding hydrogens to a structure or by slightly changing VDW radii (with the vdwdefine command) -- but you have to do this before any surfaces fail because after a failure surfacing won't work again until you restart Chimera. You create a surface with Actions->Surface->show or the "surf" command. If with experience you know that surfacing certain models is problematic, you can restrict the surfacing to particular models by selecting them before using the menus or providing the model numbers as arguments to the surf command (e.g. "surf #0,3-7,9"). Thirdly, can chimera calculate electrostatic charge on amino acids surrounding a binding cavity and reflect that charge onto the closed cavity surface? Can this surface be included in overlay view? You can use the Add Charge tool (in the Structure Editing category) to add partial charges to all atoms. You can then use the Render By Attribute tool (Structure Analysis category) to color the surface and/or atoms based on the partial charges. For more precise electrostatics, you can use an external tool such as APBS or Delphi to generate an electrostatic grid, and open that in Chimera and then use the Electrostatic Surface Coloring tool (Surface/Binding Analysis category) to color the surface with the grid data. You can certainly show the surface(s) with the ligands overlaid. You would probably only want to show a few surfaces at most simultaneously for understandability (see next answer). Given the number of structures, I would much appreciate any advice on how to make this manageable. If Chimera cannot do the calculations, can it import multiple structures saved by another capable software, then overlay them using MultiAlign (including the surfaces)? Depending on the speed of your computer, its amount of memory and its graphics card, working with 36 structures and their surfaces at once may or may not be satisfactory (and unless you have a real high-end machine my guess would be not). You will want to use the Model Panel (under Favorites) to control which surfaces/structures are being shown -- particularly which surfaces since I imagine that looking at more than a few would be incredibly confusing. Another thing that will help is to delete parts of structures that you aren't interested in. One trick here is to use MultAlign Viewer to select one residue in each structure (create a region by mouse dragging that covers one column in the alignment -- those residues will be selected), then select the corresponding chains by clicking into the main window and hitting the up-arrow key, then right-arrow to invert the selection, and finally Actions->Atoms/Bonds->delete to delete all atoms that are not part of chains in the alignment. Hopefully your ligands will have the same chain ID as the chain in the alignment and therefore will not be deleted. You might want to look at your ligands after you've inverted the selection and make sure none are selected for the following deletion. If some are selected, deselect them with "~select ligand". As a quasi-extreme measure you could prune away parts of the structure more than a certain distance from the ligands. Select the ligand clump and use Select->Zone... to select residues further than a certain distance from the clump and then delete the selected residues. --Eric Eric Pettersen UCSF Computer Graphics Lab pett(a)cgl.ucsf.edu http://www.cgl.ucsf.edu

1 0

render several models by attribute on an absolute scale
by E. Merkley 03 Apr '07

03 Apr '07

Hello Chimera users, I am using the "Define Attribute" and "Render by Attribute" features to color structures by number of native contacts observed during a molecular dynamics simulation. I would like to compare several simulations at once, applying a single color scale to several (~6) models at once. It seems that this can't be done using a separate attribute file for each model: The "render by attribute" dialog doesn't let you change the limits past the min and max of the data in the attribute file, but each simulation (and hence each attribute file) has a different range. Any suggestions? Thanks, Eric

2 1

viewing & comparison of multiple overlay structures
by Roberta King 03 Apr '07

03 Apr '07

Hello, I am relatively new to Chimera (and the user group), but am using it to compare ligand binding pockets for multiple (36) structures of related proteins. My proteins overlay nicely based on alignment, but I am having difficulty with the comparisons of multiple structures at once. First, After overlaying multiple protein (pdb) structures with ligands using MultiAlign based on alignment, how does one easily view only the overlaid ligands? [Note: I have up to 36 structures overlaid, each with 0-2 ligands.] Secondly, can chimera then calculate a solvent-accessible closed surface for the protein cavity surrounding each ligand? Alternatively, I could calculate the cavity first, then overlay based on alignment. The goal is to view the overlaid cavities for visual comparison. Thirdly, can chimera calculate electrostatic charge on amino acids surrounding a binding cavity and reflect that charge onto the closed cavity surface? Can this surface be included in overlay view? Given the number of structures, I would much appreciate any advice on how to make this manageable. If Chimera cannot do the calculations, can it import multiple structures saved by another capable software, then overlay them using MultiAlign (including the surfaces)? I appreciate any help provided. Roberta ******************************** Roberta S. King, Ph.D. Assistant Professor Department of Biomedical and Pharmaceutical Sciences College of Pharmacy University of Rhode Island

2 1

error parsing ATOM record when coordinate long-ish
by Miguel Ortiz-Lombardia 02 Apr '07

02 Apr '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dear all, I can't read a particular atom line that looks like: ATOM 4 CA GLY A 4 -32.686-186.362 -1.715 1.00 20.00 C I presume that the reason is that the y-coordinate forms a continuous string with the x-coordinate. However, this line complies with the PDB format; I quote from the RCBS site: 39 - 46 Real(8.3) y Orthogonal coordinates for Y in Angstroms I realize that this may be seldom a problem, but I think it should be tackled. In my case, this is a dummy atom to represent a dipole moment vector with quite a big magnitude. I have the same problem in the last and the previous snapshot versions of chimera (both in linux and MacOSX) Thank you in advance, Miguel - -- Miguel Ortiz Lombardía Centro de Investigaciones Oncológicas C/ Melchor Fernández Almagro, 3 28029 Madrid, Spain Tel. +34 912 246 900 Fax. +34 912 246 976 e-mail: molatwork(a)yahoo.es - ---------------------------------------------------------------------- Et ainsi ne pouvant faire que ce qui est juste fût fort, on a fait que ce qui est fort fût juste. Blaise Pascal, Pensées -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (Darwin) iD8DBQFGD2QuF6oOrDvhbQIRApSSAJ94MwNKKsPowzYE/k61g9Yen1PArwCfSnWJ gUXmMlHVsrRuLscR3IWg/tQ= =6/U1 -----END PGP SIGNATURE-----

2 1

error parsing ATOM record when coordinate long-ish (II)
by Miguel Ortiz-Lombardia 01 Apr '07

01 Apr '07

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi again, I forgot to mention that I solved my particular problem by using a bild object to depict the vector. Just thought that the parsing of the ATOM record should be fixed. Cheers, Miguel - -- Miguel Ortiz Lombardía Centro de Investigaciones Oncológicas C/ Melchor Fernández Almagro, 3 28029 Madrid, Spain Tel. +34 912 246 900 Fax. +34 912 246 976 e-mail: molatwork(a)yahoo.es - ---------------------------------------------------------------------- Et ainsi ne pouvant faire que ce qui est juste fût fort, on a fait que ce qui est fort fût juste. Blaise Pascal, Pensées -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (Darwin) iD8DBQFGD2u7F6oOrDvhbQIRAvWJAJ9UiNDLaAv9yDbf6HPoNyhHVaGAmwCfd1SF ZHM9NIY3QX5mjFVqNdvvTQk= =Gbf0 -----END PGP SIGNATURE-----

1 0

Chimera-users April 2007