I have done homology modelling and then refined the model with MD in gromacs. Then I did clustering of trajectories and selected the average structure of the model. Finally I did energy minimization of the model. But I still have 3 residues in unfavorable regions of the ramachandran plot. All 3 residues are in middle of loops (10-15 residues in length). What is a rational explanation for these outliers ?
Hi Atul
It's quite likely that Ramachandran outliers are simply errors although you should always check if there are outliers in the templates that are genuine, unusual features. If that's not the case then you need to remodel the loops in question. In general, MD with regular force fields is more likely to worsen your model than improve. I would go back to the previous loop modelling step and try Modeller's specialised loop modelling protocol to generate a large number of candidate conformations, screening them for structures with better geometry. If your loops are very large insertions then Modeller will always struggle and you might want to consider hybrid methods like Robetta or I-TASSER. Being fragment-based, they will generally have good geometry although picking out a near-native conformation is still hard. All you can do is try a bunch of validation methods and see if they agree on one particular model.
In response to your second post, there is no reason why outliers in loops are any more acceptable than outliers in general. What you may have in mind is the fact that outliers in *crystal structures* may be more common in loops in loops since the *electron density* for those regions is often worse due to flexibility etc.
Dan
Hello.
Which residues? If you are talking about prolines or glycines, they frequently appear as outliers in the Ramachandran plot. In addition, loops are very mobile structures, and its usual that some residues can be calculated as outliers.
Best regards.
No they are not prolines and glycines, can you give a reference or published paper mentioning such outliers which are valid in loops ?
Hello again.
Please try:
(1) Lovell, S. C.; Davis, I. W.; Arendall, W. B.; de Bakker, P. I. W.; Word, J. M.; Prisant, M. G.; Richardson, J. S.; Richardson, D. C. Structure validation by Calpha geometry: phi,psi and Cbeta deviation. Proteins 2003, 50, 437–50.
(2) Davis, I. W.; Leaver-Fay, A.; Chen, V. B.; Block, J. N.; Kapral, G. J.; Wang, X.; Murray, L. W.; III, W. B. A.; Snoeyink, J.; Richardson, J. S.; Richardson, D. C. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res 2007, 35, W375–W383.
Best regards.
Hi Atul
It's quite likely that Ramachandran outliers are simply errors although you should always check if there are outliers in the templates that are genuine, unusual features. If that's not the case then you need to remodel the loops in question. In general, MD with regular force fields is more likely to worsen your model than improve. I would go back to the previous loop modelling step and try Modeller's specialised loop modelling protocol to generate a large number of candidate conformations, screening them for structures with better geometry. If your loops are very large insertions then Modeller will always struggle and you might want to consider hybrid methods like Robetta or I-TASSER. Being fragment-based, they will generally have good geometry although picking out a near-native conformation is still hard. All you can do is try a bunch of validation methods and see if they agree on one particular model.
In response to your second post, there is no reason why outliers in loops are any more acceptable than outliers in general. What you may have in mind is the fact that outliers in *crystal structures* may be more common in loops in loops since the *electron density* for those regions is often worse due to flexibility etc.
Dan
Dear Atul,
you may first check the properties of the underlying reference crystal structure you are relying on. If this amino acid is well resolved and also is an outlier in the ramachandran plot, it is a feature of this fold, not a bug.
If the loop is diordered or involved in crystal contacts in the reference structure (Check PISA-server at ebi),
your model is not reliable in this region anyway.
If loops are flexible, a single snapshot or an average will tell you nothing, you have to check the distribution to find the equlibrium angle.
Happy modeling,
Marco
One should probably ask if a couple of outliers in a loop region are worth all the effort of spending a lot of time using different modeling methods and protocols. Unless of course you are interested in this particular region for some functional reasons. But again, as people noted, without an experimental structure you would never know if the modeled conformation is correct.
Yes the protein model is partially disordered having large loops (more than 10 aa) and the template (crystal struc.) is also the same with large loops. I have attached a ramachandran plot with outliers marks.
All the above suggestions regarding the question are appropriate and provide enough information regarding quarry, but I am still eager to know about the name of residues in loop region.
They are Lysine, Glutamic acid, and tyrosine, all of them lie in the loops (more than 10 residues)
I will suggest you to go for energy minimization of protein using any software..I have done using Gromacs and took the minimum Potential energy frame to evaluate once again in RAMACHANDRAN Plot or ERRAT.
Hi, the problem is where are located these residues. If residues are in N or C terminal tail and are not relevant for the binding site, you can use the protein after the comparison with template(s). But if the residues are involved in a binding site or close to it you must remodel the loop (Prime, schrodinger is able to model loops, but there are also web-resources able to perform loop modeling) or you can rebuild the full length protein by using other templates if possible, or combining templates using multiple template based alignment. This method can improve the quality of the model and overcome your problem.
Sincerely
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology