See, I've been thinking about this and on my tight budget without funding I can't afford advanced commercial software, though I do have access to an efficient Linux computer and a university computational cluster to submit computationally heavy jobs. I've been considering that I could use Visual Molecular Dynamics (VMD), NAMD, Gromacs, and SigmaPlot in particular. I may think of other packages, though unfortunately I can't use Amber16. Sadly, I've seen a lot of great research employing Amber, though I intend to use it sometime in the future.
I recently read what I consider to be a great paper at least in leading me in the right direction for calculating free-energy of binding in protein complexes for a point mutation. I was thinking that I could follow the protocol as I get started on scripting my own molecular dynamics (MD) calculations.
Anyway, this is the paper I was talking about:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4160123/
I think it might be helpful to attempt to employ the MD method because the calculations appear to statistically approximate the results from the Isothermal Titration Calorimetery (ITC) experiments and NMR analysis in the same paper, showing great harmony between the simulation and lab experiments.
Say, what would you do if you had just the sequence of two proteins and you wanted to approximate the physics of the binding event in the absence of X-Ray Crystallography studies giving researchers Protein Data Bank (PDB) files? If you didn't know the structure of your protein, could you guess it by solvating a linear sequence of your protein in a cubic box of water and letting it fold into its lowest energy state by the hydrophobic collapse effect, for instance? (Maybe it would be be better to use a sphere of water as the solvent?)
Finally, thank you so much for helping me. If you have any alternative methods I would love to know them. Before I get too involved in trouble-shooting my experiment, I would like to know if I am on the right track.
P.S.
I was thinking about studying the effect of a point-mutation in a histone binding event. Sorry if I'm vague in explaining my goal. I'm a new scientist.
If you don't have the structure of the protein, you can't really do anything. Trying to simulate the totally unbiased folding of a protein is not practical. Only small protein domains have ever been folded, and often times those take massive computational resources; for a full-length protein you're talking about several million atoms, and you'd have to do simulations for microseconds or milliseconds. Even with the best MD software and hardware, that simulation could take years and you may not even find a real solution.
Relative binding free energies are routinely calculated by MD methods, but the prerequisite is a high-quality protein structure, some knowledge of the binding mode, and well converged simulations.
You may simply study it by enthalpy, entropy or free energy changes upon the mutation. It does not need to make the problem more complex.
Q molecular dynamics (MD) software package is especially suited for MD simulations coupled with free energy calculations: linear interaction energy (LIE) , linear response approximation (LRA), free energy perturbation (FEP) and empirical valence bond (EVB): xray.bmc.uu.se/~aqwww/q/
I think you can obtain the academic license to use Q for free.
---
For an example of the use of Q (LIE and FEP) for the calculation of mutational effects, see Klvana et al. (2012):
http://www.ncbi.nlm.nih.gov/pubmed/?term=23013478
If you don't have the structure of the protein, you can't really do anything. Trying to simulate the totally unbiased folding of a protein is not practical. Only small protein domains have ever been folded, and often times those take massive computational resources; for a full-length protein you're talking about several million atoms, and you'd have to do simulations for microseconds or milliseconds. Even with the best MD software and hardware, that simulation could take years and you may not even find a real solution.
Relative binding free energies are routinely calculated by MD methods, but the prerequisite is a high-quality protein structure, some knowledge of the binding mode, and well converged simulations.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics