Question: Molecular Dynamics Uncertainties
gravatar for Elmo
7.3 years ago by
Elmo130 wrote:

Dear all,

I'm new to Molecular Dynamics and I have some doubts about the following things that I have done:

  1. I am studying an enzyme which is not in the PDB site. So I performed homology modeling using Modeler. I modeled two structure using two approaches: 1) single-template 2) multiple templates generated from Modeler. (Q: I didn't introduce any disulphide bonds or any additional info, just based on the Modeler output, is it ok?)

  2. I have used SAVS to evaluate both the models. For the single-template approach, i obtained 91.2 core, 6.6 allow, 1.0 generous, 1.2 disallowed (Ramachandran Plot) and 89.1 core, 9.4 allow, 0.8 generous, 0.6 disallowed for the multiple-template approach. (Q: can I conclude that the second model is the better one?)

  3. Next, I tried to perform energy minimization using GROMACS. Then I did MD simulation for 10ns, but the RMSD and radius of gyration for both the models did not converge (keeps increasing).

  4. Next step I'm suppose to do is Docking this enzyme to a ligand. Can I proceed with the structure (second model)?

Please advice. Any suggestions are very much appreciated. Thank you in advance!

molecular homology modeling • 2.5k views
ADD COMMENTlink modified 4.7 years ago by Biostar ♦♦ 20 • written 7.3 years ago by Elmo130
gravatar for Pappu
7.3 years ago by
Pappu1.9k wrote:

What do you want to do with the enzyme? You can also post your query in gromacs user mailing list.

  1. I recommend you to use I-Tasser to generate models and compare to your own.
  2. Pcons is good for evaluating homology models
  3. Do your protein contain mostly loops? Which force field are you using for MD? You might consider running 50-100 ns simulation.
  4. You can use both structures for docking and compare the binding sites.

Look at my previous posts for more details. Good luck.

ADD COMMENTlink written 7.3 years ago by Pappu1.9k

Thanks Pappu for your kind reply.

My main aim is to predict the 3D structure of this enzyme and to use the enzyme for molecular docking. 1. I have used I-Tasser to generate the models and by comparing, do you mean that I do a superposition on my protein (generated from Modeller) ? 2. Thanks, I will look into that :) 3. Well, I used Jpred to predict the SS, and yes, they are mostly loops. The same with the predicted structure with Modeler. I used Gromos96 53a6 force field. I didn't run it for longer ns because of my lousy 8 core, 2.0Ghz computer which takes me forever :( 4. Can I proceed docking with the after EM structure? Or the ones after MD? (Sorry, I'm not very clear about this)

Thank you in advance!

ADD REPLYlink written 7.3 years ago by Elmo130

You have to understand the limitations of the homology modeling. Does your protein belong to the same family like the template? What is the sequence identity? For docking, you need a precise and accurate structure to start with. Otherwise all the results will be misleading. Do you have any experimental information where the ligand might bind (i.e. some interacting residues)?

ADD REPLYlink written 7.3 years ago by Pappu1.9k

Yes, my protein belongs to the same family with the template. Well, I compared it with an existing structure of the enzyme, but different organism, indicating where the active sites are and then I did a MSA and that is where I supposed the ligand might bind in my protein structure :/ I am not very sure how to proceed from here. Please advice. Thanks, Pappu!

ADD REPLYlink written 7.3 years ago by Elmo130

Just dock the ligand using a web server like swissdock or autodock and check the results..

ADD REPLYlink written 7.3 years ago by Pappu1.9k
gravatar for quentin.delettre
7.3 years ago by
quentin.delettre430 wrote:

The first point is interesting. You didn't introduce disulphide bonds... so i suppose you checked that functional and conserved disulfides bonds for your protein family have been set by modeller ?

There is a lot of server to generate homology models (Atome2, others like I-Tasser) ...and compare with automated ones already generated. Check the PMP or Modbase for example. A list exist at the Click2drug website.

If your MD didn't prove the quality of your model, it might be a problem with your protocol of MD or your protocol/structure for homology modeling. Try to do MD with the closest template structure to see could tell you which one is wrong.

ADD COMMENTlink written 7.3 years ago by quentin.delettre430

Thanks @quentin delettre for your kind reply.

Yes, I just use the templates (same functional and conserved DS bonds) generated from modeller to obtain my structure. Is it not recommended? I have tried using I-Tasser to generate my structure too. By comparing with the automated ones, do you mean that I do a superposition on both the models? I followed the tutorial provided in Gromacs for the MD simulation. However, I only run it for 10ns because of my lousy laptop which takes me forever :( Can I just proceed with docking if I were to skip MD?

Thank you in advance!

ADD REPLYlink written 7.3 years ago by Elmo130
gravatar for João Rodrigues
7.3 years ago by
João Rodrigues2.5k
Stanford University, U
João Rodrigues2.5k wrote:

MODELLER derives the SS bond information from the template, but you should always check if there are some that you know that should NOT be there and others that should.

How homologous are your templates? Do you have plenty of unaligned regions? Would it be reasonable to focus only on the active site?

Since you've already been given a lot of information for the homology modelling, which indeed is a critical step, I'll leave you with a link to an MD tutorial. How are you doing your equilibration? Which forcefield are you using? In vacuum or in water?

ADD COMMENTlink written 7.3 years ago by João Rodrigues2.5k

Thanks @Joao Rodrigues for your kind reply!

Do you mean that I'm supposed to check the SS bond from the template based on the Jpred prediction of SS for comparison? The homologous of the templates are high and not many of unaligned regions. Yes, indeed I'm trying to focus only on the active site, but I'm not very sure how. For instances, it is shown in I-tasser that the active sites are at residue 200, 214, 320. Using gromacs, do I delete all the residue from 200 - 320 to perform position restraints? I did my equilibration using 100ps for both NVT and NPT, gromos96 53a6, in water.

Please advice. Can I proceed to docking with my structure for docking, and then perform MD?

Thank you in advance!

ADD REPLYlink written 7.3 years ago by Elmo130

How much homologous exactly? If it is very homologous, then MD will not do much to it to be honest, except maybe deviate it from the template, which you might not want.. I would go straight from the MODELLER models. Otherwise, you can take a few snapshots of your MD and use it to do ensemble docking.

What kind of ligand are you docking? Do you have any hints on the active site?

Forget I-TASSER, trust your biology. What do you know of the enzyme? Where is the active site? It that active site aligned in the template?

ADD REPLYlink written 7.3 years ago by João Rodrigues2.5k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 948 users visited in the last hour