Hello everyone,
I want to do molecular dynamics simulations on single amino acids and check their torsion angles over time. Before I can even start with the simulation, I'm already facing two problems.
Is it possible to have just one amino acid in a .pdb file? Right now I was going the unsophisticated way of simply deleting every amino acid except the first one from a already existing .pdb file, but I'm afraid that this will corrupt the .pdb file. I'm basically cutting into a peptide bond on the c-terminal end. Can I somehow include the missing hydroxy group? Are there any other problems arising due to the manipulation?
Is it possible to calculate torsion angles for just one amino acid? I was able to calculate torsion angels for whole proteins with biopython but I'm missing the phi angle on my first and the psi angle on my last amino acid. Reducing the .pdb file down to only one amino acid leaves me with no angle at all. I read that you can use the H atom coming from the nitrogen to calculate phi and the O from the hydroxy group to calculate psi. If I use the narrowed down .pdb file I'm missing at least one of those atoms as they are involved in the peptide bonds.
These are very specific questions, but I just can't find a good solution.
I'm happy for every piece of advice!
Josh
Yes it is possible. The "PDB file" is just what you see when you open it in a text editor. There is no invisible "meta-information" flying around, i.e. you can't corrupt the file per se by deleting residues/atoms (I am phrasing it that way because that was exactly my fear when I was dealing with PDBs back in the day.) That said, some tools might require certain things to be present (or atoms in a certain order) and you might up with issues such as bonds not visualised properly etc.
Potentially with a tool that allows molecular editing (SYBYL, Chimera?) -- or you do the calculations manually/computationally and place the group yourself...
As I wrote above, it depends on the software.
If my biophysical knowledge is not completely rusty, I'd say no. If you can't use dipeptides/tripeptides, see if the approach you describe above is a good enough proxy. You could test how much the angles differ by using that approach on full protein backbones, where you have both the adjacent amino acids and the respective H-atom/O-atoms available.
Rather than cut down an existing PDB, I'd suggest you just use something like the
PeptideBuilderinUCSF Chimeraand then export as a PDB (or whatever). It will ensure you're getting a valid file written.Thank you both for your reply, that was extremely helpful! I will probably try the narrowed down file and the method mentioned by Joe below. I still have to figure out how to treat the torsion angels but in case I find no solution, I guess I can always work with trimeres.