Entering edit mode
7.4 years ago
l.souza
▴
80
Hello everybody,
I predicted the tertiary structure of a protein with Swiss-Model and Phyre2. Then, I refined the structure using ModRefiner. When I got a Ramachandran plot of the structure some outliers were present, like in the picture above:
The residues in the general plot are GLN; LYS; SER; GLN; and the ones in the pre-proline plot is GLN.
What is the better explanation for this outliers and how can I solve them?
I appreciate any help!
Have you looked at any other metrics for these simulations? RMSD to a known structure, that kind of thing?
I can't speak as to why those specific residues have come up as outliers, but you should just bear in mind that these are simulations, and they are rarely, if ever, 100% accurate.
I imagine what has happened with the threading approach that Phyre uses (I'm not familiar with Swiss-Model), is that there are residues the software has probably 'coerced' in the threading approach, whereby they are connecing 2 elements of secondary structure or similar, and their bond angles aren't quite right.
Have you eye-balled the structure to spot these residues and see if they look reasonable?
I have eye-balled it; one of them is in the middle of a strain and the others are in coils... They don't seem to be impossible! I haven't looked other metrics yet. Do you have any suggestion of tools to RMSD?
I use UCSF Chimera for almost everything. If you open the structure(s) used to model yours, and then your PDB, you can use the MatchMaker tool.
Take a look at this tutorial (there are too many steps to go in to here): https://www.cgl.ucsf.edu/chimera/docs/UsersGuide/tutorials/alignments.html
There's a webserver here that uses the TM-align algorithm, alternatively. From what I've read this is supposed to be a slightly more robust statistic (your mileage may vary) http://zhanglab.ccmb.med.umich.edu/TM-align/
Can see the picture! Check and upload again!
I fixed the picture... Can you see now?