I want to calculate binding free energy of antigen-antibody complex after 30ns MD simulation.
I used MMPBSA but it takes approx 12 days for 1 ns (500 frames) of antigen-antibody complex in which i also performed computational alanine scaning (2 amino residues were mutated) . so now i want to used any other software that can calculate binding free energy of 30ns of antigen-antibody complex in less time.
It is a really hard task Sidra.
In first instance using a MM-GBSA approach is less computational demanding than MM-PBSA approach (without loss of relevant information, in most of the cases). Furthermore, if you have combined an energy decomposition analysis with you MM-PBSA calculation it require additional time. However adopting a trajectory protocol (as you have done) is very time consuming, in any case.
In my experience, from the trajectories I have produced, I usually sample few representative frames (e.g.: the medoids derived from a cluster analysis); then I submit them to MM-GBSA calculation.
Finally you must keep in mind that, for various reasons, you cannot use the absolute value of calculated free energy of binding but you can use it only for comparative tasks.
Did you run it in your own workstation? Big system simulation is also slow. AMBER has a detailed tutorial to guide people how to use MMPBSA for binding energy calculation at http://ambermd.org/tutorials/advanced/tutorial3/py_script/.
Afterwards, you can also try my ISABEL and/or related tools. It launches MM-GBSA calculation from the AMBER MMPBSA.py program, performs Energy Decomposition analysis and it returns the interaction energy matrix.
https://github.com/dcorrada/RICEDDARIO/wiki/ISABEL
Dario and Xiaoquan,
I am also facing related problem in calculating MMPBSA calculations using AMBER, I was able to successfully complete MMGBSA calculations, but when PBSA started their always a error showing that calculations for complex have problems in executing sander. I hope you can suggest some solution for this problem.
Dario and Xiaoquan,
Actually i am using $GMXPBSAHOME/gmxpbsa.sh. it's GMXPBSA because i performed my antigen antibody simulation using gromos43a1 forcefield.
is it possible to use amber MMPBSA instead of gmx? and if we can use it so what will the possible difference can come by using it.
Sidra,
I am currently using AMBERTOOL 14 for MMPBSA calculations, its not difficult, rather better than gromacs mmpbsa, you just need to follow the instructions provided in tutorials. I had performed MD using NAMD and converted few of the files to AMBER format, although used original trajectories of MD I got from NAMD as input in MMPBSA.py for the BE calculations.
As I stated before, I do not submit a whole trajectory to MM-GBSA calculation.
In the protocol currently I use the snapshots are sampled from a trajectory in PDB format. Afterwards, each pdb file is minimized with sander by means of amber ff. Then, the minimized structures will be submitted to MMPBSA.py script.
Otherwise, you can produce a trajectory in AMBER and submit it to MMPBSA.py. But, as I stated before, it could be really time and computational demanding.
Did you use the whole antibody? It can be much faster if you just use the variable domain of the antibody. As for the antigen, make it as simple as possible, if possible, use the smallest domain possible that contain the epitope. And I agree with Shaqa, fixing/restraining bunch of atoms may help too, I suppose every MD tool has that feature.
I don't agree with Muhammad, a wide literature suggests to use (if possible) the structure of Fab fragments instead of only variable VL and VH domains. Furthermore, don't use any restraint if possible: it introduces a bias if you want to evaluate the dGbinding.
Dear Dario,
Yes I agree with the restraint part. But what about the truncation of the antibody? This is alanine scanning, so I think the most important part is seeing and compare the importance of certain amino acid inside the antigen binding-site. And could you please provide some literature example about the importance of the whole antibody structure rather than just the Fab fragment in molecular dynamics? I've been trying to look for it with Google Scholar, but couldn't find any. On the other hand I found that the use of Fab fragment can still give result that conform with the experimental result:
http://www.sciencedirect.com/science/article/pii/S0022283698916849
http://www.sciencedirect.com/science/article/pii/S0022283601951034
In my country (Indonesia), which is a developing country, resources are limited. So yeah, sometimes we have to compromise accuracy at certain level as long as it is acceptable. There is even a joke in here, "A good thesis is the accomplished one (or you might say doable)".
Regards,
Radif
Update: I just realize that I was mistaken previously, by variable domain, what I mean is the Fab fragment, I'm sorry Dario, my bad!
hello Muhammad Radifar,
i used only the Fab fragment of antibody and antigenic epitope.
- Badges
- Science topic