Hi Vasiliki,
Dear Sam,
I used your email from the 'Public Forum' in this link: https://simtk.org/forums/memberlist.php?mode=email&u=482
It would be very useful if I could speedup the calculations as my system is a big one (it is a 180nt RNA). I will go back to my protocol (.dat) and trying to find out which keywords control degrees of freedom and oscillations.
See whether you can reduce the flexibility by rigidifying parts of your system that don't need to move.. use "mobilizer Rigid". Maybe solve one piece of the system at a time, changing the flexible zone accordingly. Look for any large forces that would create sharp jumps at the first reporting interval. If the simulation progresses too slowly, you can set reportingInterval very small, just so you can see what is happening.
Just one more question. There is the possibility to incorporate counterions and/or water in the calculations?
You can use the monoAtoms and waterDroplet commands. They are in the reference guide. However there are a few caveats: 1. monoAtoms in 2.13 can instantiate the ions, but not turn on their PARM99 force field terms using "Physics where you want it." You may be able to turn them on if all molecules in your system have the force field terms turned on. In 2.14 this is corrected, but 2.14 has not been released yet, you would have to compile, or if you're on Mac I could send you a pre-release compiled version. 2. You have to be careful about atom names, chain ID, and residue type in your input pdb, in order to read in the ion coordiantes from a file. For instance, for Magnesium the atom name is "Mg+2", the residue type is "MG", and the chain ID has to be different from that of any biopolymers in your system. 3. water molecule positions cannot be read from a PDB file. You would create a new water droplet every time you run. 4. Most importantly, we haven't properly benchmarked water and ion performance. There are some problems with water -- I've seen close contacts between water and biopolymer kick a backbone through a proline cycle, for instance. So you're welcome to use it, but at the moment I would not compare the results to those you would get from MD. Thanks Sam
Many thanks Vasiliki
Quoting Samuel Flores <samuelfloresc@gmail.com>:
kalispera Vasili,
What forum did you post your question on? I would like to answer on there but I can't find it.
To answer, MMB is currently still a single-processor code. You will unfortunately get no speedup from using a quad core machine.
I can perhaps provide advice on how to speed up your calculation though. You will want to reduce degrees of freedom, and make sure you have no fast oscillations in your system that might force a small time step size. Let me know if you want help with this.
Sam
Hello sam,
The following is an e-mail sent to you by fadoulog via your account on "Simtk.org Forums". If this message is spam, contains abusive or other comments you find offensive please contact the webmaster of the board at the following address:
Simtk.org Forums <noreply@simtk.org>
Include this full e-mail (particularly the headers). Please note that the reply address to this e-mail has been set to that of fadoulog.
Message sent to you follows ~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dear Dr Flores,
I use MMB 2.13 in a quad-core computer (8GB RAM) and Ubuntu 13.04 64bit for folding an RNA. However, the job uses just one of the four available cores. I am wondering if I could share it to all four of them.
Many thanks. Vasiliki
-- Vasiliki E. Fadouloglou, PhD Marie Curie Fellow IMBB/Forth and Democritus University of Thrace Tel +306973969995 email fadoulog@imbb.forth.gr, fadoulog@mbg.duth.gr
---------------------------------------------------------------- This message was sent using IMBB - WebMail.