I'm trying to build 3d model of protein with modeller, unfortunately I can find no template with good identity. So If I use that template anyway, can I use any tool to refine that model?
I am not sure about the meaning of the term "refinement" in your question. Usually, refinements means the final step of the model building that includes elimination of close contacts, loop relaxation, etc. For this, you only need to run energy minimization and a short MD run. This step is necessary independently of how high is the sequence identity with the structural template.
If by refinement you mean a procedure that can get the model closer to the native structure, then there is no obvious solution. If the structure is not too large, you can try a long MD run (hundreds of nanoseconds) in hope that it will do the job. However, you will not know whether MD resulted in any improvements.
You can try to do multiple alignment to increase both identity and reliability of modeling.
Hi all, thanks for your answers. Can you please suggest me any tools/tutorials for MD simulation???
Gromacs is free, fast and easy to use for MD simulation. You can find appropriate tutorial in its website.
Hi Ruhshan,
Firstly, I agree with Krishnedun and samira. You can also validate the model by checking the ramachandran plot. You can also do that by confirming with the active site amino acids. The most reliable method to refine homolgy models is to use Prime tool which uses both the energy based (PrimeSTA) and knowledge based (ClustalW) methods.
Thank you!
Perfect tutorials for Gromacs are available here:
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/
If your protein binds any ligand, you can create a set of models and perform docking of the ligand. Then, the model with best scoring or presenting contacts that are in line with experimental data can be picked up for MD for further refinement.
You can also try an old "FRaskenstein's monster" method:
http://www.ncbi.nlm.nih.gov/pubmed/14579325
https://www.genesilico.pl/meta2/
you can use this online tool:
http://sysbio.rnet.missouri.edu/3Drefine/
Hello Ruhshan Ahmed Abir
Protein model refinement is very essential for bringing them closer towards experimental accuracy for further computational studies. Based on the research papers that have been published based on this work as well as on what most of the researchers prefer for their in silico works, i would suggest you to go for a couple of best servers for refining protein models. One is Modrefiner, an algorithm for atomic-level, high-resolution protein structure refinement; and the other one is GalaxyWEB, a web server based on the method tested in CASP9. In case, you would use a server like trRosetta for modelling, you dont have to refine the models, because the models generated come refined and energy-minimized by default. The links for all the three servers are given below:
Modrefiner:
https://zhanglab.ccmb.med.umich.edu/ModRefiner/
GalaxyWEB:
http://galaxy.seoklab.org/
trRosetta
https://yanglab.nankai.edu.cn/trRosetta/
Hope it helps.
I have used the server http://sysbio.rnet.missouri.edu/3Drefine/
mentioned by Fatemeh Ghafouri and found the significant improvement in models stereochemical properties when compared with initial model using PROCHECK and ProSA web serves.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python