Protein modelling is a complex matter. May I ask how good you background in structural biology is?

There are mainly tow different approaches.

I. Homology modelling, which uses existing modell, with (weak) homology to your protein to model your sequence. The two best I guess are:

(1) PHYRE2 (http://goo.gl/lEMBd) - it will make everything for you, it is actually made for people with less background in structural biology.

(2) If you need more control swiss model (http://swissmodel.expasy.org) could be a better choice.

If you working on a new fold, things get more complicated. This real ab-inito folding is still under active research. AFAIK is rosetta the best program for it (https://www.rosettacommons.org). However, depending on your background in linux it might be difficult to install it. Furthermore on a single computer it may run for ages (normally you run it on a cluster!). You can use the free service (http://robetta.bakerlab.org). However, last time I checked it took severlal month to give you a solution.

Regardless how you arrive at your model, the most problematic part will be to confirm that it is actuall worth the time you spend to make it. You need good prediction, which you can biochemically test to put some beef into it.

If you have nothing to do with structural biology, I would recommend to look for a collaboration with a specialised group...

Regards and good luck!

Jan

Dear Jan,

Thanks very much. I have already used Phyre2 and Swiss model but they didn't give me the good results as the good template is not available which covers whole target sequence. Phyre2 gave the model which was having only 20% residues with more than 90% accuracy. This is too low. Then I looked for ab-initio modeling. I have also submitted the sequence to robetta, let's see how much time it will take. I will also try to install Rosetta. Do I need to install whole software and also is it commercial? Please help me in this.

Thanks

Best regards,

Dear Archit,

rosetta is free to academic users. You have to apply for a license and download the whole source code. However, you have to manually build (compile) the whole software suite on your own. After installing, you have to generate a fragment library, calculating models, and cluster them appropiately.

Last time I checked (~4 years ago) the whole process was not very userfriedly and I wouldn't recommend it to newcomers.

You also have to consider, that calculating the models takes a long time. Potentially you would need a cluster, if you have a medium size protein.

Finally, a modell is always a model and you never know ho good it is, untils somone solved it experimentally. Are you really dependent on a modell that you are willing to invest a lot of time in rosetta?

Best,

Jan

Dear Jan,

I am not really dependent on the model. Actually we are planning to build a model as of now and then solve it experimentally later and then will compare both of them. Let's see how it will go. Anyways thanks. I will try to install it and will ask you if I face any problem in installing and using it. Hope you will help me.

Thanks

Best Regards,

If there isn´t good templates, you must use threading (no homology modelling). The best web server for me was I-Tasser (http://zhanglab.ccmb.med.umich.edu/I-TASSER/). Another one useful is LOMETS (http://zhanglab.ccmb.med.umich.edu/LOMETS/)

Actually threading also did not give me the good model, so I am planning to do ab-initio modeling. Please tell me which server or software would you suggest?

http://zhanglab.ccmb.med.umich.edu/QUARK/

Dear Fernando,

Quark is good for protein having less than or equal to 200 residues but my protein consists of more residues than 200. Can you please suggest me some other software?

You could use Rosetta, but is not a web server, you must download the program, and is not very user-friendly

https://www.rosettacommons.org/guide/Ab+Initio+Structure+Prediction

Thanks Fernando!! Is there any other web server other than Quark which can model the protein with more than 400 residues ab-initio?

I have found one for Rosseta (http://robetta.bakerlab.org/) and other one (http://mordred.bioc.cam.ac.uk/~rapper/). You coul try them. Please let me know how they works

good luck¡¡

Is Rapper a online server?? I think it's a software. Please tell me how it works if you have any idea.

Thanks!!

I have not ever used them. Sorry¡¡

To be honest, I don't think a "good" ab initio modeler for protein structure exists. If you don't have a homologous protein with a known structure that you could plug into SWISS-MODEL or something equivalent you are really not going to be able to trust anything you come up with. I spent several years doing protein crystallography, and something that could ab-initio predict protein structure would obviate that whole field, or at least make phasing a lot simpler as you could phase with a model. The truth is this never works (and rarely even for a homology model) because none of them are reliable.

Thanks Brandon!! I agree with you but anyways I would give it a try!!

You have I-TASSER, also by the Zhang group: http://zhanglab.ccmb.med.umich.edu/I-TASSER/. It is a mix between homology and ab initio modelling, and it works with relatively big proteins (like yours). It preformed very well on the last CASP competitions.

Dear Archit,

In my opinion, there is nothing like best ab-initio protein modeling. Though people are using online servers like Phyre2 etc but the output is not quite reliable. One is supposed to go with protein modeling only if there is sufficient similarity between the protein and its templates. In fact, I have started with the same but landed nowhere.      

Thanks for all your kind information.

From my view, I would suggest you to try using I-TASSER and If your protein will be transmembrane or GPCR, pls, try specifically, GPCR-TASSER (http://zhanglab.ccmb.med.umich.edu/GPCR-I-TASSER/). Another online protein model server (http://ps2.life.nctu.edu.tw/index.php) creating good validated model

For the best approach always to predict the low identity sequence threading and ab inito methods suitable. I have tried Phyre2, it is not given a correct model and now manually recruiting protein from template analysis for MODELLER_modeweb server (https://modbase.compbio.ucsf.edu/modweb/). In which, Modeller Key is required.

regards,

Rajesh

I want to purchase some protein modeling and docking software. Please suggest few best softwares. 

@pratap Shankar. There are many. I use Gold. It is genetic algorithm based, allowing flexible ligand, and partial flexibility for protein. Others are Glide (exhaustive search based docking program from Schrödinger), and MOE from Chemical computing group. 

What are the most powerful software for the prediction of epitopes and the modeling of the 3D structure of proteins ?

If your sequence is less than 200 AA, you can try to use this online server

https://zhanglab.ccmb.med.umich.edu/QUARK/

You can refer to below attached paper full-text, which has all the methodology detials and link to webservers.

By applying quantum mechanics of small protein chain, to calculate some theoretical physical parameters.

The question is pretty old, but here are my favorites:

I-TASSER

Phyre2

RaptorX

Other servers for contact prediction are also available.

quark works well

For ab-inito model (lower sequence similarity with pdb structures) , I think no one is perfect. But you can give a comparative perfection of model using simulation analysis. I think you can use Robetta,I-tasser,trRosetta,RaptorX.

For ab-inito model (lower sequence similarity with pdb structures) , I think use modeller ( Iterative Modeling )