June 2010 - modeller_usage

questions on loopmodel in the presence of a ligand
by Diane Lynch DLLYNCH 02 Jul '10

02 Jul '10

2 1

Cross Species Dimer
by Fuhrman, Kit 02 Jul '10

02 Jul '10

Dear Modellers, I am looking for some insight into modeling a heterodimeric complex in a single species. The original crystal structure was solved with Chain A and Chain B from two different species. Chain A is from humans and Chain B is from mice. I would like to model the whole complex in the mouse form. Chain A is about 77% conserved between mice and humans. Here's my basic protocol. 1) I edited my sequences to contain only the portions modeled in the PDB. 2) I used align2d to create my PIR and PAP files 3) I used automodel and subclassed it to only select chain A for modeling The final structure looks ok visually. The DOPE score per residue actually decreased by about .01 for a few regions between the crystallographically solved structure and final homology model. Otherwise is looks about the same. Does this protocol sound correct? Is there anything else I should have done? or other potential problems I should look out for? My main concern is that the dimer interface be as close to the real deal as possible. I ran one of the models through PDBsum and a couple of Gamma turns were lost. One turn had the same sequence in both human and mouse, but no did not show up as having the secondary structure. Any help would be greatly appreciated. Best regards, Kit Kit Fuhrman IDP Graduate Student Department of Pathology, Immunology and Laboratory Medicine University of Florida College of Medicine kit.fuhrman(a)pathology.ufl.edu

2 1

structure refinement and loop optimization protocol
by Thomas Evangelidis 24 Jun '10

24 Jun '10

Dear Modellers, I've read previous posts on the same topic and concluded that it is better to generate multiple models with moderate refinement and loop optimization level, rather that a few with very thorough parameterization. I've also noticed myself that with the thorough parameterization parts of the secondary structure are distorted. I have concluded about the optimum alignment after a lot of experimentation and would like to set up a very effective optimization process. However I'm not sure about the output files. My code looks like this: a = MyLoopModel(env, alnfile=alignment, > knowns=known_templates, > assess_methods=(assess.DOPEHR,assess.normalized_dope), > sequence='target') > a.starting_model = 1 > a.ending_model = 2 > # Normal VTFM model optimization: > a.library_schedule = autosched.normal > a.max_var_iterations = 200 ## 200 by default > # Very thorough MD model optimization: > a.md_level = refine.slow > a.repeat_optimization = 1 > > a.loop.starting_model = 1 # First loop model > a.loop.ending_model = 5 # Last loop model > a.loop.md_level = refine.slow # Loop model refinement > level > Which generates the following pdb files: target.B99990001.pdb target.B99990002.pdb target.BL00040002.pdb > target.IL00000001.pdb target.IL00000002.pdb > I thought the above should perform model refinement twice and write 5 different conformations (loop optimization) for each. So my questions are the following: 1) Can you explain what's happening with the .pdb files? 2) I 'd like to ask your opinion about the most effective way to find a near-native protein conformation in low sequence identity levels. How should the parameters shown above be set? I don't care if it's running a day or so as long as I get good results. 3) I also attempted to cluster the models with a.cluster(cluster_cut=1.5), which generated a representative structure with the parts of the protein that remained similar in most of the models but without the variable parts (files cluster.ini and cluster.opt). Does it make sense to select the model that is closer to that consensus structure? If yes is there a way to do it with Modeller? I know it can been found with Maxcluster program. Or alternatively, do you reckon it is better to select the based model based on the normalized DOPE z-score? Hope to get some answers on these question cause I've been strangling to find the best refinement/optimization protocol for several weeks. thanks, Thomas

2 3

wrong asparagine and glutamine side chains
by David Rodríguez 24 Jun '10

24 Jun '10

>From time to time I find in generated models wrong asparagine and glutamine side chains (see attached PDB for a glutamine example), either with automodel or loopmodel classes. It is not reproducible, it happens in certain runs, and within the same one it only happens in one residue of one model. I've been looking for it to be documented, but I found nothing related. In order to overcome this, I used the "mutate_model.py" script, to try the trick of mutating glutamine for itself, but didn't work. I wonder if anyone know a way to detect and/or correct those residues with a Modeller script (I know other programs that can do it, but using an exclusively Modeller pipeline will make things much easier). Thanks in advance, -- David Rodríguez Díaz, PhD Student Fundación Pública Galega de Medicina Xenómica (SERGAS) E-mail: david.rodriguez.diaz at usc dot es

2 1

Optimization and modelling
by Piyush Diyora 24 Jun '10

24 Jun '10

I followed the steps mentioned by John W. "(To find a native like structure I created 100 models for 5 different template combinations with and without loop modelling and with and without MD refinement, and I found that no loop modelling and no MD refinement gave me the best models. I took my 5-10 best models as ranked by dope and the objective function followed by PROsa analysis. Then I further analyzed the 5 or so best via procheck. IMO brute force and sheer numbers are the best way to determine emperically what combinations of settings will work best for you. I would definitely try no loop model and no MD and always do 100 models minimum. Cheers.") - John W., After choosing the best models, based on the gap (where no template structure was available), i performed slow loop optimisation. As I am new to modelling, I don't know how can I optimise my model more accurately http://sites.google.com/site/piyushdiyora/ The above link has an image before and after optimisation. Should i optimize other parts of he model, because there are several high peaks. Second question is, I am working on Dos1 protein from Fission Yeast. Sequence blast in Protein data bank provided 75 results. Out of these none of them were completely identical, but only part (fragment around 80 - 120 amino acid long) of the Dos1 sequence was identical to the (fragment) reference templates. Similarly other part was identical to other reference template. Based on such analysis, I went through all 75 protein 1) extracted the part of the sequence from reference template and then aligned with the Dos1 protein (http://sites.google.com/site/piyushdiyora/home/second-page) My question is 1) As these fragments are from different protein having different 3d structure, HOW RELIABLE WILL BE MY PROTEIN 3D STRUCTURE? 2) If I want to build a model based on the reference template (created by joining all the fragments), how should i do it? I dont know how to manually modify the sequence file (which will be used by modeller) Thank you in advance :)

2 1

Loop Refinement - Clustering necessary?
by Jan H. Löhr 22 Jun '10

22 Jun '10

Dear Modeller users, according to posts in this mailing list as well as the background information to ModLoop, the best loop-model is chosen by lowest pseudo-energy score (http://modbase.compbio.ucsf.edu/modloop/ - Fiser's and Sali's papers cited at the bottom of the page). However, the tutorial of Modeller indicates that "it is important to note that a most accurate approach to loop refinement requires the modeling of hundreds of independent conformations and their clustering to select the most representative structures of the loop" http://www.salilab.org/modeller/tutorial/advanced.html) I have been comparing different loop-models generated by loop.model for a selected region of a pdb-file and I am tempted to simply choose the best DOPE-HR-scoring model. Yet the clustering idea does makes sense. So far, the greatest cluster of models often contains (one of) the best scoring model(s), but not in every case. My question is therefore: Should the best model be chosen, or should the best model of the greatest cluster be chosen? I wonder about your opinions regarding this issue. In case anyone is voting for the clustering method: What method is easily suitable for clustering - unfortunately, the loop.model-class does not seem to have an integrated clustering option, does it? Regards Jan Jan H. Löhr Univ. Hamburg, Germany

2 1

Using my own initial structure to model a loop
by Thomas Evangelidis 16 Jun '10

16 Jun '10

I was looking at : http://www.salilab.org/modeller/manual/node26.html#SECTION:initialmodel and was wondering if it's right to pass a loop fragment generated de novo into automodel with the inifile parameter. Is it the same as using it as a template in the alignment (that's what I do so far)? If yes, can I define multiple infiles to model multiple loops of my protein whilst doing homology modeling for the rest of it?

2 1

Modeling with multiple templates
by sdh 16 Jun '10

16 Jun '10

Hello all! I want to model a big fibrous protein using several templates. The templates do not overlap (gap is no longer then 1-3 residues) or overlap over few residues. I found the following sentence in the FAQ: "If no additional information is available about the relative orientation of the two domains the resulting model will probably have an incorrect relative orientation of the two domains when the overlap between A and B is non-existing or short. To obtain satisfactory relative orientation of modeled domains in such cases, orient the two template structures appropriately before the modeling." My questions are: how precisely the templates have to be oriented? Would be cursory placing enough? How can I order the templates by using some simple constraints instead of orienting the templates? Thanks for help, sdh

2 1

Re: [modeller_usage] modeller_usage Digest, Vol 9, Issue 88
by John W 15 Jun '10

15 Jun '10

I can't address your own results, but I can tell you my experience: To find a native like structure I created 100 models for 5 different template combinations with and without loop modelling and with and without MD refinement, and I found that no loop modelling and no MD refinement gave me the best models. I took my 5-10 best models as ranked by dope and the objective function and then submitted them to PROsa. Then I further analyzed the 5 or so best via procheck. IMO brute force and sheer numbers are the best way to determine emperically what combinations of settings will work best for you. I would definitely try no loop model and no MD and always do 100 models minimum. Cheers. --- On Tue, 6/15/10, modeller_usage-request(a)salilab.org <modeller_usage-request(a)salilab.org> wrote: From: modeller_usage-request(a)salilab.org <modeller_usage-request(a)salilab.org> Subject: modeller_usage Digest, Vol 9, Issue 88 To: modeller_usage(a)salilab.org Date: Tuesday, June 15, 2010, 8:25 AM Send modeller_usage mailing list submissions to modeller_usage(a)salilab.org To subscribe or unsubscribe via the World Wide Web, visit https://salilab.org/mailman/listinfo/modeller_usage or, via email, send a message with subject or body 'help' to modeller_usage-request(a)salilab.org You can reach the person managing the list at modeller_usage-owner(a)salilab.org When replying, please edit your Subject line so it is more specific than "Re: Contents of modeller_usage digest..." Today's Topics: 1. structure refinement and loop optimization protocol (Thomas Evangelidis) 2. Using my own initial structure to model a loop (Thomas Evangelidis) ---------------------------------------------------------------------- Message: 1 Date: Tue, 15 Jun 2010 02:29:37 +0100 From: Thomas Evangelidis <tevang3(a)gmail.com> Subject: [modeller_usage] structure refinement and loop optimization protocol To: modeller_usage(a)salilab.org Message-ID: <AANLkTil1lD1Yf9cWtV5O1ID9kxv7QZkzaDcz3b4n0iBo(a)mail.gmail.com> Content-Type: text/plain; charset="iso-8859-1" Dear Modellers, I've read previous posts on the same topic and concluded that it is better to generate multiple models with moderate refinement and loop optimization level, rather that a few with very thorough parameterization. I've also noticed myself that with the thorough parameterization parts of the secondary structure are distorted. I have concluded about the optimum alignment after a lot of experimentation and would like to set up a very effective optimization process. However I'm not sure about the output files. My code looks like this: a = MyLoopModel(env, alnfile=alignment, > knowns=known_templates, > assess_methods=(assess.DOPEHR,assess.normalized_dope), > sequence='target') > a.starting_model = 1 > a.ending_model = 2 > # Normal VTFM model optimization: > a.library_schedule = autosched.normal > a.max_var_iterations = 200 ## 200 by default > # Very thorough MD model optimization: > a.md_level = refine.slow > a.repeat_optimization = 1 > > a.loop.starting_model = 1 # First loop model > a.loop.ending_model = 5 # Last loop model > a.loop.md_level = refine.slow # Loop model refinement > level > Which generates the following pdb files: target.B99990001.pdb target.B99990002.pdb target.BL00040002.pdb > target.IL00000001.pdb target.IL00000002.pdb > I thought the above should perform model refinement twice and write 5 different conformations (loop optimization) for each. So my questions are the following: 1) Can you explain what's happening with the .pdb files? 2) I 'd like to ask your opinion about the most effective way to find a near-native protein conformation in low sequence identity levels. How should the parameters shown above be set? I don't care if it's running a day or so as long as I get good results. 3) I also attempted to cluster the models with a.cluster(cluster_cut=1.5), which generated a representative structure with the parts of the protein that remained similar in most of the models but without the variable parts (files cluster.ini and cluster.opt). Does it make sense to select the model that is closer to that consensus structure? If yes is there a way to do it with Modeller? I know it can been found with Maxcluster program. Or alternatively, do you reckon it is better to select the based model based on the normalized DOPE z-score? Hope to get some answers on these question cause I've been strangling to find the best refinement/optimization protocol for several weeks. thanks, Thomas

1 0

structural alignment of models respect template(s)
by David Rodríguez 12 Jun '10

12 Jun '10

Hi, I am interested in making a 3D structural alignment of the resulting models with their template(s). I have included the "a.final_malign3d = True" line in the script as stated in the manual, and the *_fit.pdb files were generated. However, none of the models are correctly aligned to the template. I wonder if anyone knows a successful way to align the models with just one modeling script. I could provide inputs and outputs if necessary. Thanks in advance. Best regards, -- David Rodríguez Díaz, PhD Student Fundación Pública Galega de Medicina Xenómica (SERGAS) E-mail: david.rodriguez.diaz at usc dot e

2 1