Hydrogen@Home

http://hydrogenathome.org/forum_thread.php?id=351
-Jack

November 13, 2008
Write to Thomas Friedman and tell him there are garage scientists working on green energy http://www.hydrogenathome.org/forum_thread.php?id=358#2689
-Jack

November 6, 2008
Disabling the workunits temporarily. In the process of integrating Quantum Espresso Chemistry app
-Jack

November 3, 2008
I finally got a working fuel cell Alkaline Fuel Cell
-Jack

October 24, 2008
Slashdot article on hydrogen producing bacteria Hydrogen-Producing Bacteria Could Provide Clean Energy
-Jack

October 8, 2008
Hydrogen@Home is developing a home experiment for hydrogen production
http://hydrogenathome.org/forum_thread.php?id=342
-Jack

September 23, 2008
Hello, I just returned last night from Paris. If anyone is interested in the presentation, you need a program that runs this file. I know VLC player works. The presentation ended prematurely. Look under 'Jack Shultz: Modeling biological hydrogen production. Slides, Video'
http://boinc.berkeley.edu/trac/wiki/WS08/WorkshopProceedings

-Jack

September 4, 2008
I am preparing to attend the Fourth Pan Galactic BOINC Conference in Grenoble, France where I will give my first presentation on the science of Hydrogen@Home. I will be in France from September 8 - September 22nd.
-Jack

August 25, 2008
The Foundation for Computational Learning and Science is performing a survey of BOINC users. The survey asks BOINC client users about their demographic data, level of education, income, project participation, level of interest, suggestions for improvement, and technical data related to their computers running BOINC. If you would like to participate the survey is here.
-Jack

August 21, 2008
I just returned from the American Chemical Society conference in Philadelphia. This was my first experience at ACS and I attended several computational chemistry symposium. There is still much for me to learn about computational chemistry but I feel that this project is on the right research track in its development with integrating docking and molecular dynamics simulations. Understanding how to apply these tools for problems relevant to biological hydrogen production is challenging. Our strategy has been building models for proton reduction catalyzed by Fe-Hydrogenase using an open source MD and finding ways to create custom force field parameters. I see now how to do this conceptually, but there are problems with this approach. There is not enough empircal data to use for these models. The structures of the iron prosthetic groups are not fully understood so this makes it challenging to predict reaction mechanisms. If we could predict the reaction mechanism, then it should be possible to design or find alternative enzymes to Fe-Hydrogenase.

Since the problem with biological water splitting hydrogen production is the interaction between evolving oxygen molecules and Fe-Hydrogenase, another approach is keeping Oxygen away from Fe-Hydrogenase. Now how can we do that? Introduce a molecule that has a higher affinity for oxygen than Fe-Hydrogenase. Once attached, this molecule will safely float away and transport the Oxygen out the cell. Where on earth would anyone find molecule like this? Every hear of Hemoglobin? Hemoglobin transports oxygen from lungs to cells in many animal species. Whether the biochemistry of Hemoglobin is compatible with plant species is an unknown, maybe we can study the properties of hemoglobin to find similar proteins. Dock oxygen molecules to these hemoglobins, then run classical md to see how they transport through the algae. Finally, we want to know whether the protein will release the the oxygen molecules, we may be able to model this with a semi-empirical QM/MM if we can get a good description of the environment near the algea's cell wall. -Jack

August 15, 2008
Esteemed BOINCers and Loyal H@H Crunchers:

Please allow me to now formally introduce myself. My name is Andrew Gillette and I have been working closely with Jack Shultz, the founder of the hydrogen project, in the development of better computational algorithms, new site design, and in collaborations with other research groups also working on photobiological hydrogen. Over the last several months, many new and exciting things have taken place. A blue chip company under non-disclosure agreement has offered to help us for free, and with no strings attached to scale up the hydrogen project with better hardware and network resources. The local fervor in the Washington DC area over photobiological hydrogen as a renewable energy source, and alternative to painfully expensive gas prices has contributed to increased interests and more people volunteering to become H@H data crunchers. As this project has evolved, we've seen more interest and enthusiasm, greater collaborations, and we've even had the opportunity to see how our computations are practical and appreciable in the real world.

In our most recent development, I Andrew Gillette had the opportunity to visit the National Renewable Energy Laboratory (NREL) in Golden, Colorado on behalf of the project and its founder, Jack Shultz. While there, key contacts were established between leading researchers of both the computational side and bench science side of the development of photobiological hydrogen. The research coming out of NREL is impressive, and the synergy between the computational biologists and the bench scientists is very illuminating. In the coming weeks, Jack and myself will be presenting at two key events; the symposium at the University of Delaware, and the BOINC conference in Grenoble, France. As these events unfold, we will continue to keep you posted on the future evolutions of the hydrogen project and the collaboration therein. Some very exciting changes are on their way, and with this in mind we are grateful for your participation and continued involvement in the hydrogen project. As it is seen from the perspective of the bench scientists, photobiological hydrogen is not even yet at 20% of its estimated efficiency. There are still physical and biochemical barriers that must be overcome in order to achieve the estimated efficiency benchmark and in order to establish photobiological hydrogen as a sustainable renewable energy resource. Even as it stands now, exciting developments are under way. So it is with this in mind that we encourage you to tell your friends about the project, participate more, post to the message boards about the project, and continue to offer suggestions towards this project's development. The outcome of the world energy crisis is uncertain, but with your help, your ideas, your participation, and your passion, we may be able to together bring humanity one step closer to a better world.

~A~

August 11, 2008
Want to know more about Hydrogen@Home? Check out our article on the United BOINC website
What is Hydrogen@Home? Brief
http://www.unitedboinc.com/news/1-latest-news/88-what-is-hydrogenhome-brief
-Jack

August 1, 2008
Here is an interesting book on Bioinspired Chemistry for Energy! http://www.nap.edu/catalog.php?record_id=12068
-Jack

July 24, 2008
With Government and Industry Support, Hydrogen Vehicles Could Reduce U.S. Oil Use and Emissions. See the recent report from the US National Research Council http://www.nationalacademies.org/morenews/20080717.html
-Jack

July 23, 2008
Recently I have developed a marketing strategy for BOINC, which should serve as practice for marketing Hydrogen@Home. Rytis setup a BOINC store in Europe and I setup one in the US. http://boinc.berkeley.edu/wiki/BOINC_logo_merchandise This merchandise is being sold with zero markup for both Rytis and myself. The company that makes the t-shirts charges for its services. Bad Penguin offered to make a postcard using the Hydrogen@Home banner. I don't know if anyone has the original file for this banner. We would like to use it. -Jack

July 9, 2008
I am planning on attending the ACS QM/MM symposium and the 3rd CMM 'Users Meet Developers' Workshop on QM/MM Simulations, held in Philadelphia. This may prove helpful in my attempt to understand Quantum Molecular modeling. http://biocore.ks.uiuc.edu/biocore/biofs/Public/CMM%20QMMM%20Workshop%202008/info/index.html -Jack

July 3, 2008
Over the past few weeks I've dedicated much more energy to professionally networking and building relationships with technical people interested in solving the World's Energy problems. Through our BOINC meetup group I met Andrew Gillette who has background in biochemistry and software developement. Andrew earned some credits in a biochemistry graduate program at Georgetown and recently transfered to Univeristy of Maryland. Together we have developed many ideas on what we could do with distributed computing. Andrew is fascinated by the science involved with Hydrogen@Home and has developement ideas of his own. I am planning to make him an administrator on our project and grant him title 'Research and Development Director'. Soon, I hope we legally formalize this relationship under a 501(c)(3) non-profit. -Jack

June 23, 2008
Just got back from the beach. I met with Eric who is involved with several BOINC projects. We had some interesting conversations about BOINC and some of the physics research. My friend Andrew and I are considering a publication dealing with the conceptual framework of Hydrogen@Home. There is a compeling reason for publishing a paper on this topic. BOINC deserves more positive attention to help it gain social acceptance among scientists. Energy is a significant problem. Through writing this article, we hope to translate the purpose of this project to the average scientist who may be dismissive of BOINC. We hope this will also promote our meetings here in Washington, DC. Scientists from many disciplines will come to learn about this technology once they see the potential. -Jack

June 16, 2008
As much as I would like to let everyone join, this is still an alpha project and any additional computers may be an impediment to developement. We already have many Windows, Linux and Macs. Friday I plan to disable account creation until our project matures. -Jack

June 12, 2008
Our next BOINC meeting 'BOINC 101' is scheduled 5:30 Thursday June 19th at the National Academies Keck building, 500 5th ST NW Washington, DC. This will be our first really formal meeting. We will have an introduction to the concept for a diverse audience. This is a free event. No refreshments. We plan to reconvene at a local pub afterwards. For details and registration http://opensource.meetup.com/84/

June 4, 2008
Organizing our next Washington DC, BOINC meeting sometime in June. We are planning on using one of the conference rooms at my office for a information session then we will have an informal meeting at a local bar near Gallery Place Chinatown. If anyone is interested in attending, please let me know. We have a Meetup group under Washington DC BOINC. -Jack

June 3, 2008
Ok so we got that bandwidth problem under control for now. I started testing JavaGenes. My first attempt running a Java Application did not work. So then I decided to read the BOINC trac documents and try creating a windows launcher using JSmooth. Our projects version 5.11 of JavaGenes worked for a test example that uses a specific class. This Java App has many classes to choose from and I would really like to figure out how I can dynamically choose classes at run-time instead of creating a launcher that uses a specific main class. This may be something obvious that I do not fully comprehend yet. Furthermore, I still need to learn how to write a parameter file for JavaGenes. Our purpose for using JavaGenes is to automatically generate molecular force field parameters for molecular dynamics simulations. My intent is to have an interface where you can enter a substrate and products for a reaction. The project will initiate a serries of workunits that will create input for another serries of workunits. After a all this work is done, we will have candidate enzymes for a reaction. -Jack

May 27, 2008
Bandwidth consumption is approaching my monthly quota due to our frequent updates following our data corruption. I also began exploring ways to mirror our download directory. I used an rsync whenever a workunit is created to synchronize a mirror. That process seems to get tied up and never replicating data fast enough. I could also try running rsync as a daemon, use another protocol or find a way to mirror. I am asking for suggestions on this. -Jack

May 24, 2008
We had another failure of a snapshot backup. I'm going to work with the hosting provider to find out why these backups are failing. Going forward it is critical to have this safety line. -Jack

May 23, 2008
The custom workunit generator is back. Feel free to submit your own workunits. -Jack

May 21, 2008
Please reattach to Hydrogen@Home. If you experience difficulty still, try deleting your host or http://hydrogenathome.org/hosts_user.php resetting general preferences -Jack

May 20, 2008
Until further notice, I have disable the AutoDock Workunit generator. I will be generating Uppercase workunits soon just so we can re-establish our project signatures are correct. -Jack

May 19, 2008
I have the project partially working. Only windows is available currently. I am limiting the interval of workunit creation to 60 seconds to reduce the frequency of downloading. You may need to reattach to the project. -Jack

May 19, 2008
We had a probelm attempting to update the security key and signatures for our project. When we reverted to the snapshot image, it was corrupted. I was able to retrieve all the data however and I am in the process of rebuilding the server. -Jack

April 23, 2008
I need someone runing OS X 10.4 (intel) to test an application. Please PM [B^S] sTrey on our project for details. -Jack

April 23, 2008
Our current mac build is only compatible with system kernel 9.2.2 which is available on Mac OS 10.5 Leopard. We are working on a multi-OS mac intel port. -Jack

April 20, 2008
Just released a mac intel platform for autodock on our project. I would appreciate some feeback if someone could generate workunits. -Jack

April 13, 2008
I came across a software tool for force field calculations called JavaGenes. This Open source package developed by NASA is free to redistribute. I will see if these force fields are compatible with Amber and other molecular dynamics packages. -Jack

April 12, 2008
We identified a potential source of our chronic workunit problems. Certain protein files contain non-standard residues such as nucleic acids. One of the MGL Tools Developers suggested this command-line prepare_receptor4.py -U nphs_lps_waters etc. This should ignore these non-standards residues. We are running workunits specifically on the previously failed protein files. -Jack

April 9, 2008
Over the last few weeks I have studied molecular force field parameters which are required to run molecular dynamices. Force field parameters supplied by Amber, like the General Amber Force Field (GAFF) do not include parameters for molecules containing Iron. If I want to run a semi-empirical QM/MM Molecular Dynamics simulation on molecules interacting with Fe-Hydrogenase, I ought to have force field parameters for Iron. There are ways to computationally derive force fields or you can find them in a database. I'm not entirely sure how to derive them. I recently got in touch with a researcher involved with R.E.D.-III http://q4md-forcefieldtools.org/RED/ Their group agreed to give me an academic license so I could better explore charge derivations. Their group operates out of Universite de Picardie - Jules Verne. -Jack

April 7, 2008
One of the example ligand molecule files I site in the Advance Workunit Generator EXAMPLE #1: Ligand: H-Cluster NSC (107241) has a corrupted Mol2 format entry in the NCI Diversity set. I contacted a friend over at NCI involved with that system. He will check out that structure file next week when he returns from the ACS conference. -Jack

April 2, 2008
I created an advanced workunit generator for AutoDock4 workunits. This provides a more powerful interface for custom docking. In addition, we have a mailer that optionally sends results to initiator, and our sample results file will temporarily provide the results in a directory. Thanks to Nicolas, we can represent these files as 3D structure models. -Jack

March 31, 2008
We had an unfortunate database mishap. My workunit generator interface was not sending tasks to hosts. I used my backup script but it did not work. When I restored to a snapshot I lost about two weeks worth of data including some new project volunteers. I your account is gone, please join again. I am terribly sorry.

March 20, 2008
If you live or work in metropolitan New York City; Washington, D.C.; or southern California, you may be eligible to test-drive an Equinox Fuel Cell vehicle, or share your opinions directly with The GM Fuel Cell Team. You must be at least 21 years of age, a U.S. citizen and have a valid driver's license. http://www.chevrolet.com/fuelcell/checkzipcode/

March 16, 2008
I moved the domain over to a new server Friday night. Still a number of things to enable. I finally got the posts rendering properly on our forums. I'm out of town to see my folks who recently returned from fishing in New Zealand. I got a bit of a head cold so I'm a bit ineffective right now.

Jack

March 10, 2008
Started building a new project server this weekend. I've been less attentive to our current test environment.

Jack

March 2, 2008
Tomorrow I am speaking at an international forum on renewable energy at American University here in Washington, DC. http://www.hydrogenathome.org/ForumRenewableEnergy.pdf

Jack

February 23, 2008
I've been silent only because I have been really busy on behalf of my project and all of BOINC. Recently we had another BOINC meeting here in Washington, DC. And yesterday me and another participant visited with Dr. Taufer at UDel who does Docking@Home. I appreciate getting to meet her and the Docking@Home staff! We are talking about having an East-Coast BOINC meeting sometime this summer if we can drum up interest. Maybe we can call it BOINC BBQ Bash. I will keep you posted. -Jack

February 3, 2008
I have been working on a couple things, Learning about Molecular Dynamcis simulations, Learning more about how to send application files through an archive and promoting BOINC through my meetup group here in Washington, DC. I recently got a fee waiver for an academic license on Amber while concurently I integrated CP2K a molecular dynamics application. MD has a high learning curve. Currently Amber is only an education tool since it requires a license and cannot be distributed. CP2K is a powerful GPL alternative with many features but little documentation. It should be possible to make computaitonally light serialize molecular simulations with CP2K. Whether my simulations have any predictive ability is yet to be seen. I am practicing distributing applications through an archive using CP2K on windows. Unfortuetly I have not been able to compile CP2K on Linux or Mac. It should have been more difficult on Windows I thought. Finally, regarding our BOINC meetup group, I had an excellent conversation with two BOINC enthusiasts and hope to see them and more people February 20th, most likely in Silver Spring, MD.

Jack

January 24, 2008 - Washington DC BOINC
We had a nice meeting Wednesday night, with three attendees. In our casual conversation we talked off the cuff about BOINC and our various interests related to science, technology and computing. Friendly chat.

Attendees Sander, Andrew, Jack

Introductions
Sander - Very interested in SETI and crunching several years, interested in GPUs and nano-tech. Not involved with developing
Andrew - Graduate student, knowledgable about computer science, physics and biochem. Crunching for Folding@Home
Jack - Developing Hydrogen@Home to identify clean hydrogen production, learning about molecular modeling, implementing molecular docking but struggling with molecular dynamics QM/MM. Skilled in integrating existing apps.

Goals
1) Tell People about BOINC
2) Develop better mechanisms of computing, new methodologies. Perhaps collaborate on an algorithim.
a) traditional molecular analysis tools such as Pymol, CP2K, Amber & Gromacs take too long to get answers
b) Find computational short-cuts
3) GPU - See what the BOINC community can do to better support GPU
4) Learn more about the feasibility in using hand helds to crunch.
5) Study the feasibility of parallel computations through BOINC

Next meeting time/location yet to be determined. Clyde's was convenient for the three of us.

Jack

January 16, 2008
It has been a few weeks since I wrote. I did not want to jinks the current configuration, which seems to work very well for our testing purposes. I removed the sample_assimilator since I don't have the space to store large volumes of useless data from pseudo random molecular docking. But I do have a mirror running thanks to some help from several people TJM, Rytis among others. I have backups running regularly. Our credits are up. I am looking at Molecular Dynamics applications such as Quantum Espresso, Amber, CP2K and GROMACS.

Kind Regards,

Jack

December 31, 2007 - Happy New Years

Happy New Years

Holiday Newsletter

1) BOINC Integration of a powerful molecular docking simulation with genetic algorithms that help make approximations on lowest energy of interactions between two molecules

2) Custom Workunit Generators that allow users to create their own molecular docking experiment

3) The ability to view results through a Java Molecular viewer

Future Landmarks

1) Send results in an email to the user

2) Create a Re-Captcha for Account creation

3) Integrate a Quantum Monte Carlo simulation Upstream of molecular docking. Thus we can preform molecular docking on hypothetical reaction intermediates.

4) Reach the funding Goals for Year 2008

5) Achieve Institutional Sponsorship

So there is still more work to be done. Possibly the deal breaker on this project is whether Monte Carlo simulations integrated with Docking can provide a reasonable method to predict catalytic interactions. Another direction, however, using Rosetta it may be possible to engineer efficient enzymes.

Best of luck to you all during the Holidays!

Kind Regards,

Jack

P.S. If you live near Washington, DC you may be interested in joining Washington, DC BOINC & Distributed Computing. Our group will meet, discuss and promote open source distributed computing, primarily BOINC. Our first meeting is a casual dinner at Clyde's of Gallery Place on January 23rd 2008. I hope to see you there!

P.P.S. I also want to thank several people at Scripps Institute. First off, thank you Sargis for helping me get python packages functional through Pyinstaller. Thanks Ruth for identifying an issue with my docking results which made it impossible for running further analysis. And finally thank you Dr. Morris for the work you have done developing AutoDock.

December 13, 2007
Finally got a JMol integrated into the site with a web launch. Just copy a link to a pdb file in the sample_results directory, open JMol, File-->Open URL, paste the link into the Java Dialog Box and click Ok.

-Jack

December 11, 2007
I was having issues running the python script write_all_complexes. According to Ruth from Scripps, their scripts assume a blank line between Success and the time estimates near the End Of File. Previously I modified the source to remove references to the computer name on one of those lines. I went back to revise my source modification and recompile. I think I have eliminated that issue. Still I have many other issues and not having an ISP at home since I moved, makes it really difficult to administer the server.

Also I am checking whether I can replicate some results produced by The DOE National Renewable Energy Labs described in a recent paper where they docked Ferridoxin to Hydrogenase. In this experiment they were modeling the last stage of the electron transfer chain where protons are reduced to form H2. I sent a pdb file with an image representing ferridoxin hydrogenase complex results from AutoDock. Their work used FTDock which is another Docking simulation.

-Jack

December 5, 2007
So I finally solved a problem I started noticing three weeks ago. My upload issue was truncating and multiple results would not send. I just added the copy file tag to the result template and it seems good now!

-Jack

December 4, 2007
Thanks to some help from TJM at Enigma, I finally got the Apache Webserver upgraded. This may help with some upload issues we are experiencing. Results getting truncated on upload thus limiting our ability to do further analysis much less illustrate the results of the docking simulation. Drawback was data loss from mid-November.

-Jack

November 27, 2007
It would be nice to view the results of our molecular docking. I've struggled to get output displayed. For some reason results got truncated. I finally got a result with all the necessary info. I will see if I can reproduce this. Also, I am testing some molecular viewer java applets to integrate into the site.

-Jack

November 7, 2007
I could use some testers for Mac PowerPC. My developement was lagging on the Macs, but now I know how to port python packages on Mac using pyinstaller.

-Jack

November 4, 2007
Our custom AutoDock workunit generator is back. NCI's chemical structure lookup is back up

http://cactus.nci.nih.gov/

-Jack

November 3, 2007
Last night I resolved the mailer. Now I have multiple threads to this developement:

1) Explore shar as a self extracting solution for linux and macs and potentially windows, I notice our linux platform is faultering. I could fix that within the confines of pyinstaller ports or try something new with shar.
2) Optionally send result file of custom workunit to initiator. Yoyo and Nicolas gave some ideas.
3) Improve the summary of results by including several additional MGLTools scripts. This is where I make the case, I need to improve the distribution method before adding more complications to the mix.
4) Integrate a molecular viewer applet to view results. This is more of a dream right now. I've seen many viewers available on RCSB.ORG but don't have a clue where to begin. If anyone has interest in doing the leg work on that and emailing some info.
5) Use monikers to querry molecules. I don't know where to begin with that but not the greatest priority.
6) Screensaver that peaks at simulation. Something that would entice more people.
7) Integrate a molecular dynamics simulation that approximates transition state structures. This is critical to the science of the project. I am floating the idea among chemists I meet.

Of course, I am still looking to improve the site in other ways, make it more professional and use a secure implementation of captcha for registration.

-Jack

October 28, 2007
Lately I've been seeing many sucessful workunits. They are only failing when I try something new such as programing error handling with custom workunits. I think it is time to try pyinstaller packages with Mac! So Mac users be ready for some errors.

-Jack

October 27, 2007
Here are three pending issues, error handling the custom workunit (I am waiting to hear about BOINCs new feature to allow workunits going to your own host), I continue to look at compressions, and do you have trouble getting your account key through the email?

-Jack

October 23, 2007
We tried our python scripts ported several ways: pyinstaller worked, cx freeze distributed as flat files had too many files to download, ported cx freeze in 7zip had logistical issues. For now pyinstaller works, but has some flaws. We will revisit distributing as archives again. Good news, I got the stats exporting! Iam also developing a method to assign a ghost result to specific hosts.

-Jack

October 14, 2007
Finally I have a form where users can submit an AutoDock Workunit based on the parameters they supply. Like the Make Workunit before, these workunits first require authentication. Check out the "Generate Custom AutoDock Workunit" page. My next step is to send the results in an email to the user.

-Jack

October 3, 2007
We finally got the Linux binaries ported. In Linux we used source mglenv.sh. That set the environment correct for running MGL Tools python scripts without needing pythonsh. Also had to make an edit to mf.py. We also had to upgrade to MGL Tools 1.4.6 for several reasons. One being the parameter flag finally allows us to disable DPF flags. Many thanks to Sargis from Scripps!

-Jack

October 3, 2007
Finally I got these python scripts ported to windows binaries. Soon I will run them in a workunit!

-Jack

September 28, 2007
Our friend over at Enigma used Pyinstaller on another project to compress python scripts into one portable exe. I will test this out tonight.

-Jack

September 27, 2007
Looks like the python scripts are portable if we use py2exe. This requires further testing but I think it will be easier to format the files on the clients.

-Jack

September 25, 2007
We resolved the issue with the input files! In the workunit template we use a tag infile_0, infile_1 ... infile_n. These specify the open file name which is the physical name of the file specified in a create_work script. My next task is to troubleshoot docking parameter file issues. Then I can get this all integrated into one script to download/format/create workunits based on user parameters!

-Jack

September 23, 2007
Thanks to some help from David and Nicolas I got a solution for sending parameters in the create_work. They are sent in the job file. The job file refers to the applications by a logical name. For example:

autogrid=autogrid_5.51_windows_intelx86.exe
autodock=autodock_5.51_windows_intelx86.exe

These applications are then copied to slot when the BOINC client downloads them. They are given the logical name autogrid & autodock. Now my next problem is trying to get the various input files in the slot directory. I am attempting the same logical naming but running into issues.

-Jack

September 21, 2007
Our latest challenge is dynamically changing workunit parameters. When a workunit is created on a server, a script can supply command line tags which are ordinarily process by the program. The problem with our configuration, AutoDock4 depends on program output from AutoGrid4. They are both run through an XML file called job.xml. Wrapper reads this file and parses a tag called Task. It runs the application listed beneath that tag and any command lines. Currently I supply the command line in that file. These job files are distributed for every platform because they need to identify the exact name of the application in the job file. Sending them on the fly would require re-writing all job files for all platforms. That would cause a whole lot of problems. Frankly, I think the best approach is re-writing Wrapper to parse the command lines and direct them to specific Tasks listed in the job.xml.
BOINC has a class called XML_PARSER that I need to read about.

-Jack

September 19, 2007
Landmarks:

1) Enable Additional Platforms for AD4: Done at least major platforms!

2) Get AutoGrid4 enabled with BOINC for all platforms: Done! See 1)

3) Develop a script for formating batches of molecular files for Work Units: Done

4) Integrate an additional molecular dynamics tool for identifying saddle point energies of substrate: Not Started

5) Open up peer review scientific methods for predicting catalytic interactions on trials of positive/negative controls. Identify areas for improvement based on these trials: Not started

We have come a long way since I opened this project in June. We have AutoDock working on almost every platform. I have moved on to the next difficult problem, downloading/formating data for generating workunits. I now have a script that does that. Problem is trying to get AutoDock to run workunits with different files. Wrapper only knows the command line supplied in the job file. Re-Writing job on the fly may cause problems. Alternatively we can redesign AG4/AD4 to accept command line parameters through a file or redesign wrapper to accept a command line, sequester the parts for AG4 from AD4 and run the applications with these sperate substrings.

-Jack

September 16, 2007
A data corruption occured yesterday so a bunch of posts and results may have disappeared. I created a new link to donations, and I just uploaded a solaris and a 64 bit linux platform. Give it a try, see if it works.

-Jack

September 13, 2007
I can't say why, but all platforms are now working! Thanks for your testing. I have one Workunit Generator now. I will continue working on a method of formating all molecular data so we can have an automated workunit with user parameters. -Jack

September 10, 2007
Some posts may have disappeared while I was making recent changes. I'm sorry about that. The Python Scripts I install required some changes to the bash startup script, that cause problems for the project daemons so in the process of troubleshooting I reverted back to the last image which I took during the weekend. Maybe one or two posts were lost. My backups need some more work eventually. -Jack

September 9, 2007
I have added the link to my prototype workunit generator. It will prompt for a password if you are not signed in. I am also working on the download/format/workunit generator with public domain data from RCSB.ORG and NCI. -Jack

September 8, 2007
Many people would like to create workunits when I'm not looking at the server. I understand that frustration. Soon I hope to allow you to create your own docking experiments using legitimate public domain data of molecular structures. I have a prototype script that runs the examples. I'm not ready to deploy that yet. There are a couple improvements I want to make: 1) Only allow project testers to run this script 2) I am developing an automated download and formating of structure data from specific databases. Currently working on the formating. -Jack

September 7, 2007
Apparently POSIX (an api used by UNIX applications) has some functions which are incompatible with ANSI C standard which is primarily used in Windows. There is no standard header in VS2005 that supports this function, at least none that comes with the install. I am trying to find if there is a similar one that I can compile with. -Jack

September 4, 2007
SIGSYS error resolved thanks to help from Janus. Apparently I was compiling the BOINC_SAMPLES in VS2005 with debug option. Once I change it to release all those error messages with wrapper apps went away. Now I am having a problem generating processes with AD4 under BOINC Client. Interestingly there is no such problem generating processes for AG4. I am trying to make certain there are no windows memory issues with AD4. I am attempting to build AD4 with VS2005 instead of cygwin. -Jack

August 28, 2007
I'm being told the memory leaks errors should not cause the boinc errors. To check on this I am running uppercase for Windows to see if it errors. It looks like the the old Uppercase works despite reporting memory leaks. Next I will try the UC2. -Jack

August 24, 2007
I've been researching issues with wrapper with autodock/autogrid on Windows and compiled a fresh version this week. Turns out this version generates memory leak errors. It turns out almost all my recent compiles of BOINC samples with MS VS2005 generate memory leaks with the exception of worker and sleeper. I've been asking for advice, and I've also been trying to research my problem with Autodock.scripps.edu. They had some site issues this week. Misery enjoys company because yesterday BOINC had some site issues too. In the mean time I've had some interesting discussions about the Science. -Jack

August 20, 2007
I am back home and trying to analyze the problems facing Windows platforms. I've been sending WUs for worker and wrapper applications to help me identify the problems. We are getting mixed messages whether the sigsys error is being caught by cygwin DLLs, Autodock or BOINC. It may be helpful to compare the normal opperation of wrapper with the bad one. -Jack

August 14, 2007
I'm still out of town but Lars recently provided some binaries for FreeBSD and Solaris. I am testing those right now. Our goal right now is enabling as many platforms as possible before moving to the next step. The problems with Windows look somewhat grim right now. We are getting mixed messages on Wrappers for Mac. When I get back I hope to look through change sets and find out if the problems H2@Home are having with Mac are issolated to this project. -Jack

August 8, 2007
I will be on hiatus for a week and a half, but Aaron will be around checking posts. -Jack

August 7, 2007
Just had success with MacOS Intel. If you have a Power PC and still have errors or otherwise please let me know. -Jack

August 5, 2007
I am testing out the AG4/AD4 Windows binaries distributed by AutoDock. However I am running into a problem I had yesterday with Directory Hiearchies not placing input files properly, so clients are having trouble downloading a set of 16 files plus application files. My work arround yesterday was not appropriate so now it does not download correctly. There are a couple of ways to handle it, possible using a zip and unzip, which I believe climate predictor uses or I can work on figuring out why the files are not getting placed in the proper hiearchy. -Jack

August 4, 2007
Finally got a wrapper workunit to process the two apps thanks to help from BOINC_Projects list members. -Jack

August 3, 2007
We are now evaluating AutoDock as a compound application with AG4 running to supply grid parameters for AD4. This will probably generate many errors at first. -Jack

August 2, 2007
Thanks to Aaron we got the parameter files for AutoGrid4 in good working order on Linux 32-bit platform. I will turn my attention to developing a compound WU that uses the output of AutoGrid4 to run AutoDock4. -Jack

August 1, 2007
Recently I compiled AutoGrid4 which is an application used by AutoDock4 to prepare parameter files for AutoDock4. It may be necessary to deploy workunits as a compound applications where output from one is used by another. Currently I am observing errors from AutoGrid. Another issue we are looking at is avoiding stderr output that reveals computer names for privacy concerns. These are fairly easy source code modifications so if you see that, let us know. Meanwhile, Aaron is looking at compiling BOINC libraries with VC++ and abandoning the pure cygwin route for Windows platforms. Good luck to him. -Jack

July 28, 2007
Several posts and credits have disappeard in the last couple days. I have been working on improvements to backup procedures. As a result we have a better understanding of how to backup the project in case of system failures. Time to recover may take 1/2 hour to an hour and during these times the project will be unavailable. -Jack

July 26, 2007
Aaron made some hacks to get BOINC source compiled in cygwin and then built AD4 in cygwin with BOINC enabled. However, the application is not running properly. You may have workunits in windows that are causing errors, just cancel these until we get these errors sorted out. Meanwhile, I am working on OS X. -Jack

July 25, 2007
I cleaned up some errors disrupting the validator, looks like it is running okay now. There are still many things we need to work on. Here is a list of goals, most of which depends on previous goals being reached. 1) Enable Additional Platforms for AD4 2) Get AutoGrid4 enabled with BOINC for all platforms 3) Develop a script for formating batches of molecular files for Work Units 4) Integrate an additional molecular dynamics tool for identifying saddle point energies of substrate 5) Open up peer review scientific methods for predicting catalytic interactions on trials of positive/negative controls. Identify areas for improvement based on these trials. -Jack

July 24, 2007
We've had our first successful workunits returned. It took quite a bit of testing to get this far, so thanks to all of you who contributed by crunching the test workunits. This means we can move on to producing project applications for different platforms. In the mean time, there are a few work units available to provide further baseline data. These WUs should run for at least 2, but less than 30 minutes. Aaron

July 23, 2007
We now have a working Linux application. We've created some work units to test input and output. If you're a Linux (32bit) user, please grab one or two work units and let us know how it went on the forums. Once we're happy with the Linux platform application, we'll begin work on the other platforms! Thanks. Aaron

July 22, 2007
Many thanks to Redhox and L'alliance Francophone for their banner which resulted from a contest, you can now see it floating at the top of project pages. Thanks Again! -Jack

July 19, 2007
Horray for Aaron, he managed to compile AutoDock4 on Cygwin. Credit should also go to two other researchers in Australia, Edward and Philip Crawford who did the original work. This is good news for H2@H for now we have a way of enabling Windows platforms for BOINC. However, we are still trying to figure out how to get the source code to recognize the BOINC api. Every platform seems to encounter linker errors. -Jack

July 13, 2007
Le demain est jour de bastille et mon anniversaire, ainsi je pense que j'irai pêche I think I will take the day off for some fishing tomorrow, or maybe I will take part in the festivities. -Jack

July 12, 2007
Aaron Pollock from Queensland Australia, has joined as a project tester, developer in training and moderator. Welcome Aaron! -Jack

July 8, 2007
Just added a cluster map of visitors. I was having trouble tucking it in next to the animated gif. Welcome suggestions -Jack

July 8, 2007
Issues with AutoDock4 likely due to compilation issues from the binary distributed by Scripps. Scripps may not have enabled the BOINC options. I am attempting to build it with the BOINC options but running into trouble calling the dependencies. -Jack

July 6, 2007
Kudos to Evandro Menezes who compiled the x86-64 platform of AutoDock. However, we are still having trouble with workunits regardless to platforms. -Jack

July 3, 2007
My good friend Jeff Tome, his host uploaded a result file that completed the simulation but the rest of the results for that workunit returned errors. This is result file ad_0702071904_nodelete_5_0. What's puzzeling me is outcome says Init (which usually means it was only sent to the user) even though it was upload back to the server. At least Jeff has shown the simulation can run to completion with BOINC. Go Jeff!!! -Jack

July 1, 2007
I finally got one AutoDock Workunit to work! 7/1/2007 1:19:48 PM|Hydrogen@Home|Computation for task ad_0701071715_nodelete_4 finished Or rather, I looked at the result file that was uploaded and see application ran and produced output. However, BOINC registers errors, so I need to find out why. -Jack

June 28, 2007
Some bugs with this Google Maps Api page. I took it off for now. -Jack

June 26, 2007
New! Google Maps Api linked to most recent result sent. -Jack

June 25, 2007
Updated PHP-GD and that fixed the Avatar issue. Homepage Link seems ok now. -Jack

June 24, 2007
Resolved download issue, config.xml said hydrogenathome.orc/download. I finally opened the project to select groups of testers. -Jack

June 23, 2007
Many people are interested in testing and I am moving to have a small group of testers for this pre-alpha project. Today I encountered some technical issues in downloading input files. This surprised me because I never had this issue before. When I overcome this problem, I will make it a major priority to opening up project for testing. -Jack

June 14, 2007
Thanks to Krunchin-Keith, North Karolina USA...Earth (yes that's Karolina with a 'K'!) He made our Favicon!

June 3, 2007
Good news, testing with Autodock3 revealed interesting energy calculations relating substrate to active sites. Scripps announced RELEASE of AutoDock 4 under GPL 2007-05-07. I am now evaluating it for integration into HYDROGEN@HOME!

May 19, 2007
Problems with FTDock. Simulation is useless for predicting interactions for the positive control Catalase and hydrogen peroxide. Looking for open source flexible docking algorithm.

May 5, 2007
Running Positive Controls for Interactions between Catalase and Hydrogen Peroxide.

May 1, 2007
Moved Names Server. Switched to Sample_Assimilator

April 29, 2007
Mac-Intel and Power PC are now working.

April 28, 2007
Fixed issue with the validator. Still having trouble with Mac platforms.

April 15, 2007
Future ambition, develope a heuristic algorithm for optimizing simulation to better predict positive controls.

April 14, 2007
Success at running workunits on Windows and Linux. Success at creating large batches of workunits for FTDock. Building Mac platform and hope to have it tested soon.

April 11, 2007
FTDock is the only simuluation currently deployed. FTDock is a rigid Protein-Protein Molecular Docking application. In this contexts we attempt to adapt the simulation to predict catalytic interactions between flexible Proteins and Substrate. FTDock was originally developed by the Biomolecular Modelling Laboratory, Imperial Cancer Research Fund. It is free software under the terms of the GNU General Public License.