In terms of how the force fields compare w.r.t structure and dynamics, rather than speed factors, there have been a few recent articles that compared a number of the more popular force fields and compared them against experiment.
Here's a couple:
doi: 10.1021/ct300323g
doi: 10.1021/ct2007814
Both AMBER(12) and CHARMM force fields can be used for protein structural biology studies. What really matter is the protein and the types of studies you want to do. And, about the packages my answer is same, but namd will provide some extra features like- graphics accelaration, CPU accelaration, scalable simulation etc. Both Gromacs and NAMD packages will provide you to choose either of the force fields for calculation.
Hello.
Adding to the above answers, CHARMM and AMBER are (in my opinion) very good force field for protein simulations, but they tend to be very rigid. GROMOS, on the other hand, can sometimes provide the flexibility needed to assess several characteristics, and is also computational less expensive.
The best way to do is test both, check the results and choose the most suitable one.
Best regards.
After doing simulation using AMBER, GROMOS and CHARMM, I observed that CHARMM takes more time for the same protein simulation compared to other force fileds
I am now using gromacs force field for the membrane protein simulation for 1 ns. with scp/e box. I want to know how to increase the cordinates of the box or/and how to change the water to other liquid for simulation. thanks in advance
Dear Dr. Abel,
For membrane proteins I have started simulation for 1 ns now I am trying for more MDS, could you tell me what could be the range for MDS for membrane protein.
Hello.
Check this paper:
http://pubs.acs.org/doi/abs/10.1021/ct300083m
Within, several MD runs were made on a system with a lipid bilayer and a membranar protein.
Best regards.
I must agree with previous opinion - MD of membranes doesn't have much sense when we're talking about 1 ns simulations.
In AMBER vs CHARMM forcefield battle - there's no winner in protein context. However, some people claim that charmm ff is better for proteins and amber for DNA simulations.
Using GPUs in MD may speed up your work, but probably you will have to play a little bit with parameters like saving data on the disk frequency - my experience shows that in this approach it's probably the most important bottleneck.
And at the end - AMBER (software, not ff) has also the GPU support :)
Hello again.
Speciffically for Pgp, and based on the murine crystallographic structure, I can say that the major structure estabilization occurs in the first 20 ns. However, this can be largely due to the presence of the POPC bilayer and the linker (absent in the crystallographic structure).
Yes, even at 60 ns significant shifts were monitored. 100 ns must be the minimum for a MD run. But will depend on the ff, the membrane type, temperature, computational facilities, etc.
In an 80-core Infiniband-linked cluster, 100 ns of a system comprising over 200.000 atoms (Pgp+lipid+water+ions) takes about 20 days. So, before any long run, be sure that your system is stable enough (so you wont loose precios time if something is wrong).
Best regards.
PS: New versions of GROMACS already have GPU support for several FF!
Dear Dr. Abel, Dr. Ferreira, Dr. Rafal I am really thankful for your advice I will surely apply to my work.
I am agree that the space rquirement is mandatory for long run MD simulation.
In terms of how the force fields compare w.r.t structure and dynamics, rather than speed factors, there have been a few recent articles that compared a number of the more popular force fields and compared them against experiment.
Here's a couple:
doi: 10.1021/ct300323g
doi: 10.1021/ct2007814
- Badges
- Science topic