- modeller_usage - salilab.org

can modeller9.14 build a protein with modified amino acids?
by Ming Tang 28 Apr '15

28 Apr '15

Dear Ben, I modified both the top.lib and restyp.lib in modlib fold to include the hydroxyproline information. The command line added to the restyp.lib is like this: HETATM | HYP | O | | HYP | hydroxyproline There are HYP residues indicated as HETATM in the template pdb file, and when I used a target sequence does not include O in the .ali file, I got the backbone. However, when I used a target sequence containing O in the .ali file, I got the following error: Then, I searched the archive: http://salilab.org/archives/modeller_usage/2010/msg00413.html There, you said Modeller cannot build models containing modified amino acids. If this is the case, how can modifying both the top.lib and restyp.lib to add the nonstandard amino acids information make any sense? Thanks in advance.

2 1

How to go about refinement of multiple loops in a protein model?
by vipul＠nccs.res.in 28 Apr '15

28 Apr '15

Dear Modeller_usage members, I have made 100 models of a ~260 residue protein (enzyme) using Modeller. The model has three loop regions which have significant deviation when compared with the template structure used for modelling. the first loop is due to an insertion in the model protein's sequence(i.e. this region of 3 residues is absent in the template sequence); second loop is a 16-residue long loop located at active site cleft (a region not resolved in template) ; third loop also doesn't have information from template but is a 4-residue loop. I have the following doubts and questions: Do I need to refine all of these three loops? and should there be an order in which to refine each of these loops ( i mean, starting from N-terminus of the larger loop, etc)? Do small loops (1-5 residues) need to be refined only by the modeller script 'model-loop-tiny.py'? If yes, please explain, why? Or the regular 'loop.py' script is fine? Thanks in advance, Vipul

2 1

restyp.lib modification error
by Ming Tang 27 Apr '15

27 Apr '15

Dear modeller users, I modified the top.lib to include the information of HYP, and added the following line in restyp.lib. HETATM | HYP | O | | HYP | hydroxyproline But when I execute the align2d.py, I got the following error message: Date and time of compilation : 2014/07/30 10:34:01 MODELLER executable type : i386-w32 Job starting time (YY/MM/DD HH:MM:SS): 2015/04/27 21:32:10 map_res_217E> Single-character residue type - is not a standard residue type (one of the first 21 residues in restyp.lib). read_re_255E> PDB residue type unrecognized: BLK Traceback (most recent call last): File "C:\Users\tangm5\Downloads\basic-example\molecule\align2d - Copy.py", line 3, in <module> env = environ() File "C:\Program Files (x86)\Modeller9.14\modlib\modeller\environ.py", line 45, in __init__ self.libs = Libraries(self._restyp_lib_file, self._rand_seed) File "C:\Program Files (x86)\Modeller9.14\modlib\modeller\libraries.py", line 21, in __init__ self.__read_libs() File "C:\Program Files (x86)\Modeller9.14\modlib\modeller\libraries.py", line 26, in __read_libs _modeller.mod_libraries_read_libs(self.modpt, fh.file_pointer) _modeller.ModellerError: read_re_255E> PDB residue type unrecognized: BLK If I only modify the top.lib, it won't come out this error and can run smoothly. Could anybody give me any suggestions about what's wrong with my modification? Thanks in advance.

2 1

DOPE score based model selection
by vipul＠nccs.res.in 20 Apr '15

20 Apr '15

Dear Modeller_usage members, Is it necessary to select the best protein model based on DOPE scores? I have made 100 models of a ~260 residue protein (enzyme) using Modeller. Then I chose the best model based on the DOPE Score and did loop refinement on a 16-residue long loop (50 refined models created) located at active site cleft. In loop-refined model the loop occludes the active site cleft, but in another loop-refined model (with intermediate DOPE score) it is alright. Can I use this loop-refined model? Also, is it essential to refine the loop even when some initial Models have good conformations of the loop? Any suggestions would be very helpful. Thanks in advance, Vipul

4 6

Error :Too many neighbors; overflow ignored:
by Stroebel D 17 Apr '15

17 Apr '15

Hi, I am currently building a model of a protein complex. My models (automodel) were working fine. Trying to improved the input (introducing missing atoms, chain breaks, new multitemplate...) I got stuck with a log file finishing like this: " make_re_417_> Restraint type to be calculated: phi-psi_binormal Dynamically allocated memory at amaxstructure [B,KiB,MiB]: 20567746 20085.689 19.615 openf___224_> Open ${MODINSTALL9v14}/modlib/mnch1.bin openf___224_> Open ${MODINSTALL9v14}/modlib/mnch1.mdt neighs__292W> Too many neighbors; overflow ignored: 51 50 neighs__292W> Too many neighbors; overflow ignored: 51 50 " ...The last line being repeated as if it was the output of an infinite loop. I guess my alignement and pdb are good because they have worked previously and if any misalignment is present, the error messages should appear before the ini file is created. My .py has nothing special. (see at the end of this message.) The program created the .ini, but not the .rsr file. The .ini files looks ok unless for one thing: I was surprised that the missing atoms in the file are still represented as 'missing' in the ini output. example: "ATOM 23557 CG GLN J2975 9998.0019998.0019998.001 0.00 99.99 C " Shouldn't the remaining unknown coordinates having been rebuild using internal coordinates from the Charmm topology library BEFORE the writing of the ini file? (it may be an unrelated issue...) (The Ini file also contains: SSBOND lines at the beginning and CONECT lines at the end) I come to you since I didn't find anything usefull to explain this error. If you have any idea where it could come from, I would be very gratefull if you could share it! Thanks in advance. David my.py : " from modeller import * from modeller.automodel import * # Load the automodel class from modeller import soap_protein_od log.verbose() # directories for input atom files env = environ() env.io.atom_files_directory = ['.','../atom_files'] class MyModel(automodel): def special_restraints(self, aln): rsr = self.restraints at = self.atoms a = MyModel(env, alnfile = 'AllAlignfinal.ali' , # alignment filename knowns = ('TMD', 'TMD2', 'TMD3', 'TMD4') , # codes of the templates sequence = 'TMDRat' , # code of the target assess_methods = (assess.DOPE, soap_protein_od.Scorer() )) a.starting_model= 1 # index of the first model a.ending_model = 3 # index of the last model # (determines how many models to calculate) a.md_level = refine.fast a.make() # do homology modelling " I let the "special restraints" definition because I may have to use it for an other modelling step, but it has no use here. (and it is not involved in my problem - tested) Please note that I give you here a multi-template model building (on multichain template), but adjusting the file to a single template gives the same error message. The "soap library" is also (of course) not involved (tested). (Sorry, I first send this message to "modeller-care", but obviously it is more a problem of "usage".)

1 1

Re: [modeller_usage] DOPE score based model selection
by Mouses Stamboulian 16 Apr '15

16 Apr '15

then driven from what score I choose the best model from a set of produced models? DOPE? GA341? mol-pdf? ________________________________ From: Min-yi Shen <minyishen(a)gmail.com> Sent: Thursday, April 16, 2015 4:57 PM To: Mouses Stamboulian Cc: vipul(a)nccs.res.in; modeller_usage(a)salilab.org Subject: Re: [modeller_usage] DOPE score based model selection Dope has the tendency to pick the most compactly packed model because it favors contacts. For disordered regions of proteins this may or may not be a good thing. On Thursday, April 16, 2015, Mouses Stamboulian <mouses.stamboulian(a)lau.edu<mailto:mouses.stamboulian@lau.edu>> wrote: the DOPE score is not a correct indication from my experience. Ive modelled proteins of already known structures, then calculated their RMSD by aligning my computed models through modeller with the structure found in PDB. it has shown that not always the model with the lowest DOPE or nomralized DOPE for that matter has the best RMSD results with the PDB structure. ________________________________________ From: modeller_usage-bounces(a)salilab.org <modeller_usage-bounces(a)salilab.org> on behalf of vipul(a)nccs.res.in <vipul(a)nccs.res.in> Sent: Thursday, April 16, 2015 4:31 PM To: modeller_usage(a)salilab.org Subject: [modeller_usage] DOPE score based model selection Dear Modeller_usage members, Is it necessary to select the best protein model based on DOPE scores? I have made 100 models of a ~260 residue protein (enzyme) using Modeller. Then I chose the best model based on the DOPE Score and did loop refinement on a 16-residue long loop (50 refined models created) located at active site cleft. In loop-refined model the loop occludes the active site cleft, but in another loop-refined model (with intermediate DOPE score) it is alright. Can I use this loop-refined model? Also, is it essential to refine the loop even when some initial Models have good conformations of the loop? Any suggestions would be very helpful. Thanks in advance, Vipul _______________________________________________ modeller_usage mailing list modeller_usage(a)salilab.org https://salilab.org/mailman/listinfo/modeller_usage _______________________________________________ modeller_usage mailing list modeller_usage(a)salilab.org https://salilab.org/mailman/listinfo/modeller_usage

1 0

how to build a collagen microfibril?
by Ming Tang 13 Apr '15

13 Apr '15

Dear all, I want to build the collagen I microfibril structure. I have the tough structure 3HR2 from PDB, which contains only the Carbon and rattus norvegicus, and the amino acid sequences of bovine. I learnt from the literature that homology modelling can be used to achieve this goal. I am totally new in Modeller. Where should I start? Is there any examples for me to follow? Thanks inadvance, Ming

2 1

Silencing all Modeller output
by David Menéndez Hurtado 25 Mar '15

25 Mar '15

Hi, I wonder if there is a way to silence all output when running automodel. I have set modeller.log.none() at the beginning of the script, but I still get a bunch of output: 0 atoms in HETATM/BLK residues constrained to protein atoms within 2.30 angstroms and protein CA atoms within 10.00 angstroms 0 atoms in residues without defined topology constrained to be rigid bodies >> Model assessment by DOPE potential DOPE score : -13973.771484 >> Model assessment by DOPE-HR potential DOPE-HR score : -7906.894043 >> Model assessment by DOPE potential DOPE score : -13973.771484 >> Normalized DOPE z score: -0.645 [...] >> Summary of successfully produced models: Filename molpdf DOPE score DOPE-HR score GA341 score Normalized DOPE score ----------------------------------------------------------------------------------------------------------- MODEL.B999910001.pdb 12623.23730 -13973.77148 -7906.89404 1.00000 -0.64460 MODEL.B999910002.pdb 12689.78906 -13755.89648 -7777.28760 1.00000 -0.54034 [...] I have also tried to redirect stdout to /dev/null from Python while Modeller runs, without effect, so I guess this must be coming from a compiled printf. Is there any other way I am overlooking? In my case I have several processes running in parallel, and all my output is cluttered by this. Thank you, David.

2 2

Is energy minimization possible in MODELLER?
by Arijit Ghosh 16 Mar '15

16 Mar '15

Hi dear MODELLER users, I made a good homology model with MODELLER, it's showing good credibility in all the programs in SAVES server, but I want to do an energy minimization before I go for docking some small ligands onto it. Please let me know if it is possible in MODELLER? Regards, Arijit -- *Arijit Ghosh* *4th Year Integrated MSc.* *School of Biological Sciences* *National Institute of Science Education and Research* *Bhubaneswar, Orissa-751005, India* *Alternate electronic contact: arijitghosh2009(a)gmail.com <arijitghosh2009(a)gmail.com>* *Handheld: +91-986-171-9393*

2 1

native overlap of pdb structures
by Mouses Stamboulian 10 Mar '15

10 Mar '15

Dear Ben, I'm working on a project similar to the work you and Andrej Sali did in 2003 (comparative protein structure modeling by iterative alignment, model building and model assessment), Except that I'm using Scatter Search Heuristics rather than Genetic Algorithms. after running my algorithm and comparing the 3D structures obtained with the ones found in PDB, the results are showing good RMSD values, however using the superpose method of Modeller: from modeller import * env = environ() env.io.atom_files_directory = [''] mdl = model(env, file='3HLA') mdl2 = model(env, file='3HLA.B99991001') aln = alignment(env, file='superpose_alignment.ali', align_codes=('3HLA', 'ABCD')) atmsel = selection(mdl).only_atom_types('CA') r = atmsel.superpose(mdl2, aln, rms_cutoff =5, reference_distance = 10, refine_local=False, superpose_refine=False) rms = r.rms drms = r.drms print("%d equivalent positions" % r.num_equiv_pos) mdl2.write(file='superpose_alignment.txt') I'm not getting such good overlaps within 5 angstroms. my question is that is there a way to refine my final comparative Model so that I'll get better (higher) native overlap values? Thank you.

2 1