Dear All,
I have two different proteins and experimental evidence suggests that the interaction mediated by them is an electrostatic component. I modelled the protein, calculated the electrostatic surface potential and found two distinct patch of charged amino acids (both). After that I performed a structure replacement docking (from a crystal structure) and performed the MD simulation in GROMACS. After 50ns MD simulation, I found the no conserved interactions between the two and literature suggests interaction exists.
In this situation what should be my approach! Whether I will perform multiple simulations or, other simulation approaches will help me understanding the said interaction? Kindly help in this regard.
There are a whole host of issues that could be the problem.
1. You are starting from a modeled protein. Perhaps its structure is wrong.
2. You initiate the simulation from a docked complex. Perhaps that docked complex is wrong.
3. 50 ns is not very long, and indeed one trajectory is usually not definitive proof of anything. A longer simulation (hundreds of ns or more) may be needed. But before you invest the time in doing this, you should back up to the beginning and make sure your starting structures have any basis in reality.
#Justin,
1. The first structure is considered for the study is a solved (X-ray) and the other one is a modelled one (40% identity with template, no Gap).
2. The docking Performed were based on existing on experimental data and a model of Protein-protein complex (solved by X-ray). I just superimposed my structures with the experimental counterparts.
3. After superimposed docking, I cleared the bumps and performed the simulation. Before that, I tried several P-P docking servers, However, I couldn't get the proper orientation.
4. After 3-5 ns of MD .. I found the structure deviates. For so .. what can be done?
Well, a plausible model built from homology modeling still isn't necessarily correct. Often times, people are comforted by the level of sequence identity, but there are weird cases for which that number means nothing :)
I'm a bit confused as to what you're describing in point 2 onward. What "model of protein-protein complex" are you using? Are you docking a different binding partner to one of the same proteins, e.g. protein A is shared by both complexes but protein B is different?
The biggest problem with static docking of proteins is the absence of induced conformational change. You can look into this by simulating each protein separately for a long period of time to see how much their structures change. This gives you an idea of how binding may modulate the structure. It may simply be a bad assumption that some crystallized dimer will be useful for building a different complex.
In point 3, I'm assuming your starting structure ('cleared the bumps" = energy minimization?) has some of the experimentally observed contacts? If not, then there is really not much hope for a short MD simulation (and 50 ns again is not a very long time) to overcome any biases in the starting structure.
What does point 4 mean? What deviations? Of course the structure will change, but does this mean you are losing experimentally observed contacts, or what?
#Justin,
#2. See. The structures [monomers (model/ x-ray structure) or, model of protein-protein complex (solved dimer) ) taken in this study sharing similar structural fold. When I am superimposing the structure, I found quite similar in fold and surface-charge distribution. So I superimposed my proteins with the solved P-P structure.
Cleared Bumps:= energy minimization!
#4. When I performed MD simulation of the complex (docked), I found monomers are not sticking to the positions.. !
I was thinking to restrain the residual positions, However I don't think this will be good idea!
- Badges
- Science topic