p41-IBUCH_1_cmEGFR_110914-0-20-RND6459_0 completed in 6h32m.

I took 36.4 hours on my GTX275 (linux)! (131166gpu seconds and 127569cpu seconds)

This was a p20_IBUCH_3_cetumutEGFR_110912-1-10-RND5186_1. Very long unit! so I lost the 24 hour bonus :-(

Hi criadoperez, thanks for completing the very long task.
Ignasi dropped everything to address the size concerns. No more extra long cetumutEGFR tasks are being released. The new cmEGFR (Cancer) tasks are much shorter, and should finish in time for the biggest bonus on your GPU.

PS. It seems my initial estimates may have been closer to the mark for the 6.14app and my altered/present estimates for the cetumutEGFR tasks might be better for the 6.15app. Not that it matters now that the task size has been reduced.

I've read about this research, and I wonder if GPUGrid can help in its progress. If your method is applicable, do you plan to (or are even allowed to) contribute to this research?

I've read about this research, and I wonder if GPUGrid can help in its progress. If your method is applicable, do you plan to (or are even allowed to) contribute to this research?

It's certainly a very nice piece of work.
In theory, one could simulate all the chemicals, the transporter, the membrane, (not the milk I am afraid) and try to visualise and measure the binding and competition process that they describe. In practice it is not feasible yet:

1) AFAIK, the structure of the transporter protein (NIS) isn't solved. This means that we don't know how its atoms are arranged in space (a requirement for molecular simulations). Alternatively, one has to build homology models which is like doing "patchwork", trying to reconstruct an unknown structure, from already known structures with which it shares a fair amount of amino-acid (subunits of proteins) sequence similarity. This, in itself is another field of study and a large amount of work unrelated to molecular simulations. Preferably, we always work with experimentally-determined (solved) structures for a start.

2) AFAIK, breast milk hasn't been really modelled yet as a solvent for molecular simulations. We should add to the existing water models all components of milk. Right now, physiologic type of simulations containing water only include sodium chloride (salt) or other ions/types of salts.

3) The timescales of the transport for the chemicals is way too slow for the computation times we can afford. The plot in this figure shows the curves (concentrations) of the chemicals in their test. If you see the timescale of the exchange is of the order of hours. Currently the slowest molecular process that we can simulate and measure using GPUGRID is of the order of several microseconds.

Personally, I don't think we are ready yet to handle such system/process using molecular dynamics simulations.

Thanks for the interest!

Thank you for your detailed answer!
Truth is that when I read the text I linked, I had the feeling right from its start that this is far beyond my understanding, so I didn't look at the figures at all.
I hope it will become possible in a few years to use MD simulations on these timescales and proteins.

WRT the structure of the NIS protein, I found a description of the Putative
(presumed) ‘secondary structure’ with this image,

Ref. http://edrv.endojournals.org/content/24/1/48.full

I could not find a 3D structure, suggesting as Ignasi said the tertiary and quaternary structure is not accurately known.
Doesn't surprise me looking at it; a 13 domain bi-lipid layer spanning protein complex. While these trans-membrane segments are probably fairly rigid themselves, they could float about somewhat independently within the bi-lipid layer as the external parts of the protein might be fairly flexible, or could form a tight quaternary structure, or both at different times/under different circumstances. While a lot is known/presumed about the secondary structure, who knows what sort of functional channel it forms.

Zoltan, I think you have glimpsed the future; the expansion of the present molecular simulation systems into new territory, with more complex fluids/solutions, and perhaps even the merging of molecular dynamics with existing and future fluid modeling techniques.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Zoltan, I think you have glimpsed the future; the expansion of the present molecular simulation systems into new territory, with more complex fluids/solutions, and perhaps even the merging of molecular dynamics with existing and future fluid modeling techniques.

Indeed.

Nice contribution skgiven.
Predicting secondary structure is pretty straightforward nowadays. But as you say, unfortunately not enough to understand its tertiary structure in the membrane. Let me emphasize that transmembrane (insoluble) proteins are typically the hardest structures to solve.

Have a nice weekend,

Doesn't surprise me looking at it; a 13 domain bi-lipid layer spanning protein complex. While these trans-membrane segments are probably fairly rigid themselves, they could float about somewhat independently within the bi-lipid layer as the external parts of the protein might be fairly flexible, or could form a tight quaternary structure, or both at different times/under different circumstances. While a lot is known/presumed about the secondary structure, who knows what sort of functional channel it forms.

Yes, Aheh, Hmm, That's what I was thinking. Aheh, exactly, hmmm, Aheh.

Seriously though interesting read.

---

Radio Caroline, the world's most famous offshore pirate radio station.
Great music since April 1964. Support Radio Caroline Team -
Radio Caroline

I've read about another interesting publication. I think it's closer to the timescales of GPUGrid.