Rosetta@home questions

Message boards : Rosetta@home Science : Rosetta@home questions

To post messages, you must log in.

AuthorMessage
Profile rochester new york
Avatar

Send message
Joined: 2 Jul 06
Posts: 2736
Credit: 1,689,612
RAC: 340
Message 89645 - Posted: 26 Sep 2018, 19:48:01 UTC

One of the members is asking me questions I'm sure someone more qualified can answer >>>> I'm wondering how the tasks are calculated and also the work of the job scheduler rosetta home

One user told me that one task to check the accuracy of the calculations is duplicated a thousand times - is it true or not?

Each computer calculates a unique protein, or repeats the calculations of other tasks?

If this is the case, then 99% of computers are wasted, because tasks are duplicated for verification and reliability. It's right? or does each computer calculate a new protein?

I propose to make the UPTIME rating of computers on the network or the time for calculating tasks, the most powerful computers must calculate unique jobs without duplication, I want to calculate a unique protein, and not repeat the calculations of top-end servers.

And one more question, intermediate save and results are sent to the rosetta home server?
ID: 89645 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator
Project administrator

Send message
Joined: 22 Aug 06
Posts: 3528
Credit: 0
RAC: 0
Message 89647 - Posted: 27 Sep 2018, 3:28:50 UTC

Most of those are about one issue. So, I'll try to cover the background.

Because Rosetta is researching new methods of computational design, it is common to test new ideas with known proteins. As such, your machine does not get a unique protein. It may run a new design method against proteins that do have known structures. Those results are needed to confirm whether the new approach is truly an improvement.

Verification of results is very easy, so there is no need to perform the same calculations on more than one machine. However, what it sounds like may have been misunderstood here is that it takes thousands of work units to study one protein. So thousands of hosts are studying the same protein, but they are not redoing the same work.

Dr. Baker explains it by suggesting you picture exploring an unmapped planet, in search of the highest point on the surface. Rosetta's approach, is basically to send thousands of explorers dispersed all across the surface, each one reporting what is observed (calculated) about the area near where they were deployed. So, to verify a result like that, all you have to do is go to the specific location that is claimed to be the highest point, and do one verification of the elevation at that location. You still might not have the absolute highest point on the planet, but if you deploy enough explorers, odds are very good that your reported and confirmed best result is very close to the highest point. Something R@h also does. in some cases, is redeploy additional explorers around the area of the initial best results. This helps assure the best result is discovered.

[quote] Are intermediate save and results are sent to the rosetta home server?[quote]

No, intermediate results are stored to the host's hard drive during checkpoints. Only when a task is completed are the results sent back to the servers.
Rosetta Moderator: Mod.Sense
ID: 89647 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile rochester new york
Avatar

Send message
Joined: 2 Jul 06
Posts: 2736
Credit: 1,689,612
RAC: 340
Message 89662 - Posted: 29 Sep 2018, 20:18:34 UTC - in response to Message 89647.  

Thanks
ID: 89662 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Rosetta@home Science : Rosetta@home questions



©2019 University of Washington
http://www.bakerlab.org