Charmm and the project

Hello,

I know Docking@home is applying Charmm to research and gain knowledge of protein-ligand interactions(Information of "ligand": http://en.wikipedia.org/wiki/Ligand_%28biochemistry%29 ).

I've thought that Charmm means Chemistry at Harvard Macromolecular Mechanics, which has been developed by several scientific groups and people at Harvard University. Are people at Utep involved in this project too? If so, how much are they involved, partly or rather mainly?

Also, I noticed that some people used the word 'CHARMm'. Does this mean the same thing as 'Charmm' does?

thanks,
suguruhirahara

____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

Hi,

The team at the Scripps Research Institute is working with Harvard University to implement new biophysics capabilities of CHARMM (Chemistry at Harvard Macromolecular Mechanics). We, at UTEP, are working on the integration of BOINC in CHARMM together with the team at the Scripps Research Instutute and the team at UC Berkeley.

CHARMM is a code developed by Harvard and other universities. To use it, it is required to have an academic license and the people who are working with this code (developers and users) are universities and research centers (non-profit organizations).

CHARMm is the commercial version of the code. To use it, it is required to have a commercial license. We are not working with the company that sells CHARMm or collaborating with them.

Best wishes,

Michela

Hello,

I know Docking@home is applying Charmm to research and gain knowledge of protein-ligand interactions(Information of "ligand": http://en.wikipedia.org/wiki/Ligand_%28biochemistry%29 ).

I've thought that Charmm means Chemistry at Harvard Macromolecular Mechanics, which has been developed by several scientific groups and people at Harvard University. Are people at Utep involved in this project too? If so, how much are they involved, partly or rather mainly?

Also, I noticed that some people used the word 'CHARMm'. Does this mean the same thing as 'Charmm' does?

thanks,
suguruhirahara

I see,
thanks for reply:)

PS charmm seems relatively slow to be improved... I wish it would be opensource software:(

I wish it would be completely rewritten ;-)

Andre

PS charmm seems relatively slow to be improved... I wish it would be opensource software:(

Hello,

How is the data being returned from BOINC/CHARMM shaping up? Is it reliable and consistent enough for the results to be used yet?

Happy Crunching!

-- David

____________
The views expressed are my own.
Facts are subject to memory error :-)
Have you read a good science fiction novel lately?

Yes definitely,

Just put in app updates for all platforms, which will hopefully bring the number of invalid results down, because of the better checkpointing. Also, we have to analyze the db a bit better for results returned after the HR upgrade of a couple of weeks back. If all of this looks good and we get the protein/ligand data set in a good shape (shouldn't be too long), we can go to beta :-)

Thanks
Andre

Hello,

How is the data being returned from BOINC/CHARMM shaping up? Is it reliable and consistent enough for the results to be used yet?

Happy Crunching!

-- David

I mentioned this project on the Wikipedia.
http://en.wikipedia.org/wiki/CHARMM

See the chapter "CHARMM and Volunteer Computing".
Revise and make this project more famous;-)

thanks,
suguruhirahara

____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

Hi everyone, I am glad that the project is moving forward. I hope that it gets back on track really soon and that I can donate some cycles (CPU and Brain if possible). I remember that during the last days of the project at UTEP we were waiting for a new charmm version can someone inform me if the new version has been delivered and tested? also what is this version number?

Hi
I work with CHARMM, and I search in the available comands the way to do a docking (emap, grid, galgor....),because in my group does not exist scripts of this kind. So, I found D@H, but how i can use it?.
Thank you.

Hi
I work with CHARMM, and I search in the available comands the way to do a docking (emap, grid, galgor....),because in my group does not exist scripts of this kind. So, I found D@H, but how i can use it?.
Thank you.

Does this thread say that Nvidia is about to release a new compiler for the right computer language to compile CHARMM for Nvidia GPU use?

http://www.gpugrid.net/forum_thread.php?id=1406

If so, you might want to contact Nvidia to clear up a few problems with their announcement:

1. The person who wrote their announcement does not seem to use the same definition of "soon" as most other people.

2. No information on whether the new compiler is only for the new GT300 graphics chips.

3. No information on whether large parts of the programs need to be manually rewritten in CUDA for a reasonable speedup.

AMD/ATI appears to be slower about releasing a suitable compiler, but has recently announced an OpenCL compiler, apparantly the first computer language for which an Nvidia compiler is also available:

http://developer.amd.com/GPU/ATISTREAMSDK/Pages/default.aspx

Another compiler to check if it's suitable:

http://www.pgroup.com/lit/whitepapers/pgicudaforug.pdf

http://www.pgroup.com/resources/cudafortran.htm

Yet another compiler to consider:

http://www-ad.fsl.noaa.gov/ac/F2C-ACC.html

Appears to require more changes to the source code, though.

An idea how to offer at least some checkpointing:

Fortran programs are generally single-thread. Therefore, if the CHARMM program ever waits for a response from your shell, it looks worth a try to simply write the entire dataspace the CHARMM program uses to a checkpoint file at such a time, then later check if copying this checkpoint file back to the dataspace for the CHARMM program restores it well enough to continue from that checkpoint.

Also, you probably have better access to the CHARMM developers than most of the public, so could you ask them if Harvard offers any online courses that include remote access to the source code for whichever CHARMM program you are using, for people who don't find it practical to come to the Harvard campus or your Udel campus. If they do, I'd consider taking such a course so that I can use my years of Fortran experience in order to have a try at converting that program to run on an Nvidia GPU, and possibly also helping with the checkpoints issue.

On my checkpointing idea: I've found that, for most compilers, you would also have to write the set of registers to the checkpoint file, and also make sure that you restore all parts starting at the same locations in memory. Even more might be needed that I haven't found yet.

I'm now taking an online class in preparation for another online class in C++, although at an online school not involved in any BOINC project. Is that enough that I could qualify to get online access to the source code under an academic license?

A question - does the program have enough sections that can be done in any order without affecting the results that converting to a GPU version would make it run significantly faster?

BOINC still barely has any support at all for GPU workunits written in OpenCL, although it can handle CPU workunits written in OpenCL. I've seen a few messages saying that this is likely to change in the 6.14.* series of versions, and also a few saying that there will be no 6.14.* series of versions; a 7.0.* series of versions is next.

On Fortran compilers: I've now found that there is a Cygwin gcc-series and a gcc4-series compiler for Windows that allow combining Fortran code with C code, C++ code, and a few other languages in the same program. They seem to be mainly translations of similar compilers for Linux.

(duplicate)

(duplicate)