#social docking

There is a volunteer computing project going on right now called “Social Docking” that is written entirely using HTML5 and is run completely within the web browser. It’s an initiative to find compounds of pharmaceutical interest towards AIDS and Alzheimers. All you have to do is visit this HTML5 app socialdocking.appspot.com and let the app be in your browser for about 10 minutes before closing it. You can open a new window to browse the web or do whatever you want on your computer during that time. The idea is to get the simulation to run for 3000 iterations. This low cost supercomputing project will help researchers study three molecules including, HIV-1 integrase, which is a promising target for an AIDS vaccine.

The advantage of socialdocking.appspot.com is that it is web-browser based (as opposed to client-based Folding@Home, SETI@Home, which use BOINC). It also integrates as a Facebook app at http://apps.facebook.com/socialdocking but ChemDoodle’s XHR level 2 doesn’t work when Facebook users have SSL enabled on their accounts. Eric Jang, the app’s creator, had to remove the WebGL-related things from ChemDoodle and modify it to use JQuery Pollen AJAX methods. Thanks to level 2 XHR to fetch chemical structures from SMILES strings + Web Workers, the virtual screening app automatically retrieves the next available job once it finishes docking one chemical. It can be run for $0 using Google App Engine’s free billing scheme. Right now, web browsers have not put a lot of effort towards implementing the subworkers specification of Web Workers because there are not many applications found for Web Workers. The hope is that cheap supercomputing projects like these will improve the attention Web Workers receives from web browsers.

Here is more on the science behind this great HTML5 app from the app’s creator Eric Jang.

Developing new drugs is super expensive. It takes a billion dollars and years of clinical trials for company like Pfizer to roll out a new drug into the market. One way that these companies save a lot of time and money is to use Virtual Screening – instead of testing every possible drug candidate in a test tube, they refine a massive database of chemicals by simulating the molecular physics that go on when a drug binds to a receptor. This is called the lock and key model, which states that if a ligand binds strongly to a receptor, it is likely to induce some kind of change with the receptor, and hopefully to some beneficial effect within the cell.

For this project, I am studying three molecules at the same time and trying to find ligands that bind strongly to them. The blue one on the left is HIV-1 integrase, which is a promising target for an AIDS vaccine because it is critical for the HIV virus integrate itself into the host’s DNA replication system. The second one is Acetylcholinesterase, which can be targeted for treating Alzheimers and other forms of dementia. Acetylcholinesterase breaks down a neurotransmitter, so I’m interested in targeting the inhibitor domain that hopefully will render Acetylcholinesterase inactive. Finally, the telomeric RNA quadruplex is a unique secondary structure that personally I have been interested in for some time because I’ve done research on the past with telomeres and G-quadruplexes. It would be really interesting to find some kind of chemical side chain that binds to the arm loops and can allow us to study its role in telomerase regulation and find out more about the topology of quadruplexes.

Please visit the Social Docking app. It might be the first practical distributed computing application that uses HTML5 Web Workers.