6
$\begingroup$

I'd appreciate a suggestion for software to compare models of two proteins in an alternating electric field. The more detail, or perhaps direction towards an existing and similar paper, the better. Right now I'm particularly interested in comparing rotation of the proteins as a result of the field. The first protein, tubulin (1TUB), will be modeled as a dimer while the second, septin (2QAG), will be modeled as a trimer.

The electric field will consist of a 200Khz field at 1.5V/m with a spatial resolution of 300Ax300Ax300A. I believe that I will only need a temporal resolution of 1.5 milliseconds for 10 solutions(?) per cycle for 3 cycles in the 200Khz field (particularly unsure of this part).

In terms of computational power I have access to two 7870 Radeon GPUs, but am also interested in cloud computing if that's not enough.

$\endgroup$
7
  • $\begingroup$ this is a tough one. Maybe you can help those, who know the field, by specifying temporal and spatial resolution, duration requirement for modeling. Molecular Dynamics approach might help, it has great resolution in t and xyz, but model duration is ~1us, as I remember. $\endgroup$ Apr 11 '15 at 18:38
  • $\begingroup$ Mathlab has many good toolboxes for many things, although it is not cheap. Good news is that there are alternatives such as Scilab that is totally free. $\endgroup$ Apr 12 '15 at 19:45
  • $\begingroup$ @aandreev I've done more searches with inclusion of "Molecular Dynamics" and have found GROMACS which, apparently unlike NAMD, allows for an alternating field. This may be the program I'm looking for... onlinelibrary.wiley.com/doi/10.1002/anie.200703987/full $\endgroup$
    – Timmons
    Apr 12 '15 at 21:18
  • $\begingroup$ @Timmons I am saying, you under-specified your problem. Are you looking for 1ns temporal resolution? Are you interested in how each atom moves? Or you can approximate your molecule as a sphere with uneven charge density? That kind of descriptions might be helpful. $\endgroup$ Apr 12 '15 at 22:16
  • $\begingroup$ @aandreev ahh, my mistake. I have not "picked" a temporal resolution, but I know that I'm not interested in each atom, rather, how the whole molecule rotates in the alternating electric field. A sphere might work, however, I'd like to eventually to add several of the proteins linked in a filament (realizing that's another level of complexity) $\endgroup$
    – Timmons
    Apr 12 '15 at 22:26
4
$\begingroup$

MD

Broadly this could be attempted with Molecular Dynamics simulations. As you discovered in your search, one of the most popular softwares is free: GROMACS. This is an atomistic approach to the problem. It also has many electrostatic parameters.

I would point out that generally MD requires careful and robust methods and often an expert lending hand if the results are to be trusted. In order to do MD at the timescales for observing rotating molecules in the field, massive amount of HPC computational time will be required. This is a full blown project to get it to an appropriate level for publication.

This is computationally intensive.

Quick Electrostatics

If a HPC is not available to you then MD may not be an option. If all you want to do is visualise the location of the charge and make qualitative predictions or visualisations, PyMol and ABPS might be a quicker combo. It is also good for quantitative analysis, but obviously is a static model.

$\endgroup$
3
  • $\begingroup$ Thanks for the additional information. I've actually just come back to this problem after finishing some other work, so your timing is serendipitous. $\endgroup$
    – Timmons
    Mar 23 '16 at 18:48
  • $\begingroup$ Your point re: the necessity of an HPC is well taken. I thought my 2 7870 Radeon GPU's might be able to handle it, but your comment suggests otherwise. Do you know whether cloud computing is a viable/affordable option? (aws.amazon.com/government-education/scientific-computing) $\endgroup$
    – Timmons
    Mar 23 '16 at 18:53
  • $\begingroup$ @Timmons I've only been a part of three academic groups that were all involved in MD, but they each had their own HPCs and clusters. Whilst I can't see anything obviously wrong with cloud computing, I would email ahead and check that what you want to do is viable. $\endgroup$
    – James
    Jun 9 '16 at 8:52

Not the answer you're looking for? Browse other questions tagged or ask your own question.