- modeller_usage - salilab.org

PDB format problem...
by 유아림 12 Oct '02

12 Oct '02

Hi Modellers! I have the pdb format problem. I have the same problem with Stephane Coillet-Matillon's error written at 19 Mar 2002. Eric Feyfant replied it was a chain identifier problem. " I corrected Stephane error. It was a chain identifier problem. The chain was specified in the pdb but not in the top file, and Modeler could not extrapolate. Eric" But I don't know how to correct this problem. How to solve this problem? Thanks, Ahrim The error messages is like this " Read the alignment from file : acc_ipn1.ali Total number of alignment positions: 375 # Code #_Res #_Segm PDB_code Name ------------------------------------------------------------------------------- 1 1QJF 331 1 1QJF 2 ACCO 314 1 ACCO unknown TOP_________> 121 107 CHECK_ALIGNMENT check_a_343_> >> BEGINNING OF COMMAND openf5__224_> Open 11 OLD SEQUENTIAL ./1QJF.pdb rdpdb___303E> No atoms were read from the specified input PDB file: 1) Possibly because an incorrect/non-existent PDB file is specif ied. 2) Possibly because the segment is specified incorrectly in the alignment file or by the TOP variable MODEL_SEGMENT. That is, the beginning residue number and/or chain id in MODEL_SEGMENT may not be found in the input PDB file; MODEL_SEGMENT: 1: To find out more, switch on maximal output by 'SET OUTPUT_CONTROL = 1 1 1 1 2' rdabrk__288W> Protein not accepted: 1 rdabrk__290E> Number of residues in the alignment and pdb files are different: 331 0 For alignment entry: 1 recover____E> ERROR_STATUS >= STOP_ON_ERROR: 1 1 " YOO, AHRIM Process System Lab. Department of Chemical Engineering, Korea University --------------------------------------------------------------------------------------- Don't believe what your eyes are telling you. All they show is limitation. Look with your understanding, find out what you already know, and you'll see the way to fly.

1 0

Atom index out of range
by Satyan Sharma 11 Oct '02

11 Oct '02

Hi Modeller users, I am trying to model a 2 domain protein. I give a pdb file with 2 chains (toatal of 1828 atoms)along with the alignment. During the run I get an error - Atom index is out of range. nselat__596E> Atom index is out of range (iatm,natm): 1829 1828 Any idea what could be wrong Cheers, Satyan

1 0

new user and domain movements
by Jenny.Carmichael＠csiro.au 10 Oct '02

10 Oct '02

Dear Modellers, I am a new user of modeller having just completed my PHD in X-ray crystallography. I have two questions about modeller. 1) Do the template positions of domains in space, for multidomain proteins, have to always be pre-arranged/aligned by me to the approximate postions of the putative model before running modeller ? I ask this because I tried to model a diabody using two identical Fv fragments (VH/VL), I input two copies of the same template pdb coordinate file for both Fvs in the diabody, aligning the respective domains to a sequence that ran: VH-linker-VL(chain break)VH-linker-VL where the linker is any thing from 0-15 residues, the linkers were loop modelled. ___________________ ___________________ | | | |N | VL |--------| VH | C |__________________| |___________________| N | | | | | VH |--------| VL | |__________________| |___________________|C linker The result was that Modeller returned some ridiculous models where the two Fvs were basically occupying the same space. I repeated this excercise by arbitrarily aligning the two Fv domains such that the C-terminal of the VH was close to the N-terminal of the VL for each linker region. This time modeller returned results that were much more sensible but the Fv domains had not moved far from the original positions in which I placed them, I would like to influence the result less ! 2) In a way this is similar to the first question. How can I model the possible linker flexibility in a diabody ? I would need to fix the atom positions in each Fv with respect to each other but not thier actual coordinates, whilst allowing a variety of linker models to be built. There are few diabody template structures available and the flexible linkers point to the potentiality of a large number of possible conformers for these molecules, I suppose what I really want to know is "Is modeller the program for this type of modelling ?" Thanks for your patience JennyC

3 2

Windows
by John Penniston 09 Oct '02

09 Oct '02

Is anyone out there trying to use Modeller under Windows? I discovered today that the directory names are too long, so that I can't access them from DOS. Of course Windows is fine with such long names, but use of DOS is necessary. I'd like to contact someone who is using Windows. -- John T. Penniston, Ph.D. Dept. of Biochemistry and Molecular Biology Guggenheim 16, Mayo Clinic 200 First St. SW Rochester, MN 55905 USA

1 0

No log file in Windows 2000
by John Penniston 07 Oct '02

07 Oct '02

I'm trying to learn to use Modeller, but am getting no log file when I run the tutorial example. It creates the following files: 1fdx.D00000001 1fdx.V99990001 1fdx.B99990001 1fdx.ini 1fdx.rsr 1fdx.sch What look like error messages appear in the DOS window while the program is running, but they are not written to a file. -- John T. Penniston, Ph.D. Dept. of Biochemistry and Molecular Biology Guggenheim 16, Mayo Clinic 200 First St. SW Rochester, MN 55905 USA

1 0

Multiple Template Alignment Problems - MAXRES
by Alex Brown 04 Oct '02

04 Oct '02

Hi. At the moment, i am going through the examples found in 'Comparative Protein Structure Modeling. Introductionand Practical Examples with Modeller' from the Methods in Molecular Biology Series. Everything seemed to be o.k., until I started the third section - using multiple templates: Using the following TOP script: SET OUTPUT_CONTROL = 1 1 1 1 1 SET ALIGN_CODES = '1hms' '1lid' SET ATOM_FILES = '1hms.pdb' '1lid.pdb' MALIGN3D SET ADD_SEQUENCE = on, ALIGN_BLOCK = NUMB_OF_SEQUENCES READ_ALIGNMENT FILE = 'blbp.seq', ALIGN_CODES = ALIGN_CODES 'blbp' ALIGN2D WRITE_ALIGNMENT FILE = 'blbp-1hms-1lid.ali', ALIGNMENT_FORMAT = 'PIR' WRITE_ALIGNMENT FILE = 'blbp-1hms-1lid.pap', ALIGNMENT_FORMAT = 'PAP' I get the following error reports in the .log file: Dynamically allocated memory at amaxseq [B,kB,MB]: 2200483 2148.909 2.099 TOP_________> 2 2 SET ALIGN_CODES = '1hms' '1lid' Dynamically allocated memory at amaxseq [B,kB,MB]: 2205269 2153.583 2.103 TOP_________> 3 3 SET ATOM_FILES = '1hms.pdb' '1lid.pdb' TOP_________> 4 4 MALIGN3D Dynamically allocated memory at amaxbnd [B,kB,MB]: 2560929 2500.907 2.442 openf5__224_> Open 11 OLD SEQUENTIAL 1hms.pdb rdpdb___223E> Array too small. Increase MAXRES current maximum, current need: 30 31 rdabrk__288W> Protein not accepted: 1 rdabrk__291E> Sequence difference between alignment and pdb : STRUCTURE RES_IND ALN_ITYP ALN_RES X_ITYP X_RES -----*----- 1 1 33 GLYP 0 -----###### rdabrk__288W> Protein not accepted: 1 openf5__224_> Open 11 OLD SEQUENTIAL 1lid.pdb rdpdb___223E> Array too small. Increase MAXRES current maximum, current need: 30 31 rdabrk__288W> Protein not accepted: 2 rdabrk__291E> Sequence difference between alignment and pdb : STRUCTURE RES_IND ALN_ITYP ALN_RES X_ITYP X_RES -----*----- 2 1 33 GLYP 0 -----###### rdabrk__288W> Protein not accepted: 2 fndatmi_285W> Number of residues <> number of atoms; atom code: 31 0 CA fndatmi_284E> Beginning atom index for a residue out of bounds: 0 237 recover____E> ERROR_STATUS >= STOP_ON_ERROR: 1 1 Why won't MODELLER read the template sequences, and how do I increase the MAXRES variable. I am running MODELLER6v2 on Darwin (Mac OSX). Many Thanks, Alex Brown.

2 3

RE: Basic Modeling Question
by michael hanlon (BITS) 04 Oct '02

04 Oct '02

Straight-off I'd go for the higher sequence id. But yes - much better to use both. Choice needs to consider what your model is for. Also what are the other features of these templates, e.g. are they complexed with anything? You should take a look at the TvLDH example using MODELLER, which would appear to be very relevant to your question. It makes an excellent tutorial - demonstrating how the first template choice whilst having a significantly beter seq. id. than other alternatives did not in the long run turn out to be so useful as there was poor resolution in the active site region. So this was then combined with another template, having better resolution in that region, though poorer overal seq. id, and the combo produced a far better final model. Mike -----Original Message----- From: Alex Brown [mailto:abrown@nimr.mrc.ac.uk] Sent: 04 October 2002 15:22 To: modeller_usage(a)salilab.org Subject: Basic Modeling Question Hi. Sorry if this question seems a bit basic, but you don't learn if you don't ask, and you may as well ask the experts. The alignment of the target sequence with a suitable template is, perhaps, the most important factor in producing a good model. However, choosing a suitable template can be a problem. There appear to be two factors : 1) Sequence identity; and 2) Template resolution. Which is more important ? For example, I have a couple of possible templates I could use : Template A : 87% identity, 2.80Å resolution Template B : 62% identity, 1.60Å resolution. Would it be better to use both ? Cheers, Alex.

1 0

Basic Modeling Question
by Alex Brown 04 Oct '02

04 Oct '02

Hi. Sorry if this question seems a bit basic, but you don't learn if you don't ask, and you may as well ask the experts. The alignment of the target sequence with a suitable template is, perhaps, the most important factor in producing a good model. However, choosing a suitable template can be a problem. There appear to be two factors : 1) Sequence identity; and 2) Template resolution. Which is more important ? For example, I have a couple of possible templates I could use : Template A : 87% identity, 2.80Å resolution Template B : 62% identity, 1.60Å resolution. Would it be better to use both ? Cheers, Alex.

1 0

SS prediction
by Santiago Ramon 03 Oct '02

03 Oct '02

Hi modellers, I was wondering if Modeller is able to predict secondary structure for a multiple sequence alignment. If so, could you indicate me how to do it? If it does not predict it, could you suggest me any other program to carry out secondary structure prediction for a list of aligned sequences? Many thanks, Santiago *********************************************** Santiago Ramón-Maiques Instituto de Biomedicina de Valencia Consejo Superior de Investigaciones Científicas C/ Jaime Roig, 11 46010 Valencia, Spain ***********************************************

1 0

surface electrostatics and hydrogens
by Douglas Kojetin 03 Oct '02

03 Oct '02

Hi All- I've developed some models using MODELLER using mostly the "default" parameters shown in the tutorials. We wanted to look at the electrostatic surfaces of the proteins, but realized we were doing so without side-chain hydrogens. I compared a two solved structures of the same protein, one from NMR and the other from x-ray crystallography -- and also stripped the NMR structure of its hydrogens. Then I used the program MOLMOL to look at the electrostatic surfaces -- resulting in a difference between models w/ and w/o hydrogens. I've got a few questions about this: 1. I think I read on a previous discussion that you can add hydrogens to models during the modeling process -- how "exact" is MODELLER at placing the hydrogens? i.e. orientation of the hydrogens, pKa estimation of side-chains, etc ... 2. Is there any way to add the hydrogens after developing the model using MODELLER (or another program) -- that would not be difficult? 3. Is it normal convention to develop models with side-chain hydrogens -- or only in certain circumstances? I have not looked much yet, but I don't think that models developed and published in other papers are submitted to the protein databank ... ? Thanks for your input, Doug Kojetin

2 1