CASP6 Farming

**MikeTimbers** · 19th July 2004, 10:21

Well this is getting a little painful.

I know the little proteins are quicker which should allow us to catch those pesky Frogs more quickly but it's a lot more work that the usual long runs.

I am harvesting the farm today so that the uploads have some scientific benefit and it looks like I have 32 million wunits to go up my little pipe tonight

Then there will be a smaller upload tomorrow night. This means a significant investment of time to zip, unzip and upload and shepherding.

I guess what I'm trying to say is that if we get an even faster one tomorrow I may have to limit the number of machines working on it

Daniel, Laura, and Nora · 19th July 2004, 17:20

Thats a pile of work no doubt. Fortunately its a labor of love.

Would it make sense for you to cut back on the DF uploads by running predictor on more of your boxes?

**MikeTimbers** · 19th July 2004, 17:58

Predictor won't run nonet and I don't want to run any DC work through the firewalls. Next week I'm on holiday Friday and Monday which means that any work done after Thursday can't gain scientific benefit. Does that mean I shouldn't leave it running? Hmm. Then I'm away for two weeks in August which means next to no work on three proteins running. This CASP6 protein rotation couldn't have come at a worse time as far as my contribution is concerned.

Daniel, Laura, and Nora · 19th July 2004, 21:21

Seti?

BennyRop · 22nd July 2004, 03:29

My main computer crashed a day into my 3 week vacation 1.75 years ago, and I was bumming for having lost all that folding time.. So I sympathize..

Guess we'll just have to do our best to keep the frogs just where they are, so we don't lose much ground while you're off having fun.

BennyRop · 9th December 2004, 04:09

I've found one of my systems still running DF with a mere 14,000 files in the directory (14k/3 gens..) - since one of the firewalls had been installed that give you the option of selecting "don't give this program access to the internet"..

Ouch.

And this silly Casp 6 meeting in Italy was over on Dec 8th.. and still no word on the results. I can't find the "what are the new approaches this year", "what are the most improved approaches this year", and how we fared with the proteins we worked on..

They live to torture me.. argh!!!

**MikeTimbers** · 9th December 2004, 12:59

I'm hoping still that some useful results come out of DF to justify the huge effort this whole team put in. It's feeling a whole lot like Genome@Home to be honest.

BennyRop · 9th December 2004, 14:57

I was talking about on the actual CASP6 homepages.. there's a few things happening just after the meeting in Italy that were related to protein prediction, and it probably takes most of day to go from Gaeta? to Rome to New York, to Ottowa. So I didn't expect instant news out of Howard, Dr. Hogue, or Elena..

**pelligrini** · 10th December 2004, 11:48

Quote:

Originally Posted by MikeTimbers

It's feeling a whole lot like Genome@Home to be honest.

The GAH project has published several times, with more to come. I don't think there's any way to say that there wasn't any useful results to come from that project.

**MikeTimbers** · 10th December 2004, 14:51

Quote:

Originally Posted by pelligrini

The GAH project has published several times, with more to come. I don't think there's any way to say that there wasn't any useful results to come from that project.

Didn't mean to denigrate the science of either project, more the feedback and communication.

BennyRop · 26th December 2004, 14:58

The results are up - and DF got some results in the middle of the pack, and some that were in the bottom quarter or lower; if I'm reading those charts right.

Some were mentioning that Predictor at Home was most likely to ace the CASP6 competition - although it's definately a very young project. Being built on BOINC, it's got some of the exact same programming problems we've dealt with - small programming team. And they, too, went down for a couple of months while rewriting their client to work with the latest? BOINC client.

**MikeTimbers** · 26th December 2004, 17:12

I quite liked Predictor but it was unbelievably unreliable at the beginning and that continues as far as I can tell.

**Bruce** · 3rd January 2005, 19:04

Quote:

Originally Posted by MikeTimbers

I quite liked Predictor but it was unbelievably unreliable at the beginning and that continues as far as I can tell.

I, too, though it had possibilities but simply had too many bugs in the first version. I hope somebody checks periodically and notifies the rest of us if they come up with a better client sometime in the future.

**MikeTimbers** · 4th January 2005, 05:32

I'll check with some people who are doing it that I know. Still obviously based on BOINC so the problem of too high a priority may well still be there.

BennyRop · 9th February 2005, 15:56

Any response from them, Mike?

**MikeTimbers** · 10th February 2005, 03:33

Seems that it's not easy to get wus so you have to buy into the whole BOINC thing supporting multiple projects to ensure you have work. The validation approach can mean that you won't get given units to crunch even if there are some unless your particular machine specification matches what P@H want.

I'm still staying away...

BennyRop · 10th February 2005, 04:13

I saw the list of this month's bug reports for PaH - and for a project that's supposed to be over 2 years old (they took part in Casp 5) - they are having quite a few major crises.. So the temptation to help out a project that did better than DF in Casp 5 evaporated quickly.

19th July 2004, 10:21	#1
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	CASP6 Farming Well this is getting a little painful. I know the little proteins are quicker which should allow us to catch those pesky Frogs more quickly but it's a lot more work that the usual long runs. I am harvesting the farm today so that the uploads have some scientific benefit and it looks like I have 32 million wunits to go up my little pipe tonight Then there will be a smaller upload tomorrow night. This means a significant investment of time to zip, unzip and upload and shepherding. I guess what I'm trying to say is that if we get an even faster one tomorrow I may have to limit the number of machines working on it __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

19th July 2004, 17:20	#2
Daniel, Laura, and Nora Junior Master Genome Join Date: Jul 2001 Location: New York City Posts: 1,673	Thats a pile of work no doubt. Fortunately its a labor of love. Would it make sense for you to cut back on the DF uploads by running predictor on more of your boxes? __________________

19th July 2004, 17:58	#3
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	Predictor won't run nonet and I don't want to run any DC work through the firewalls. Next week I'm on holiday Friday and Monday which means that any work done after Thursday can't gain scientific benefit. Does that mean I shouldn't leave it running? Hmm. Then I'm away for two weeks in August which means next to no work on three proteins running. This CASP6 protein rotation couldn't have come at a worse time as far as my contribution is concerned. __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

19th July 2004, 21:21	#4
Daniel, Laura, and Nora Junior Master Genome Join Date: Jul 2001 Location: New York City Posts: 1,673	Seti? __________________

9th December 2004, 04:09	#6
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	I've found one of my systems still running DF with a mere 14,000 files in the directory (14k/3 gens..) - since one of the firewalls had been installed that give you the option of selecting "don't give this program access to the internet".. Ouch. And this silly Casp 6 meeting in Italy was over on Dec 8th.. and still no word on the results. I can't find the "what are the new approaches this year", "what are the most improved approaches this year", and how we fared with the proteins we worked on.. They live to torture me.. argh!!! __________________

22nd July 2004, 03:29	#5
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	My main computer crashed a day into my 3 week vacation 1.75 years ago, and I was bumming for having lost all that folding time.. So I sympathize.. Guess we'll just have to do our best to keep the frogs just where they are, so we don't lose much ground while you're off having fun.

9th December 2004, 12:59	#7
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	I'm hoping still that some useful results come out of DF to justify the huge effort this whole team put in. It's feeling a whole lot like Genome@Home to be honest. __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

9th December 2004, 14:57	#8
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	I was talking about on the actual CASP6 homepages.. there's a few things happening just after the meeting in Italy that were related to protein prediction, and it probably takes most of day to go from Gaeta? to Rome to New York, to Ottowa. So I didn't expect instant news out of Howard, Dr. Hogue, or Elena.. __________________

26th December 2004, 14:58	#11
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	The results are up - and DF got some results in the middle of the pack, and some that were in the bottom quarter or lower; if I'm reading those charts right. Some were mentioning that Predictor at Home was most likely to ace the CASP6 competition - although it's definately a very young project. Being built on BOINC, it's got some of the exact same programming problems we've dealt with - small programming team. And they, too, went down for a couple of months while rewriting their client to work with the latest? BOINC client. __________________

26th December 2004, 17:12	#12
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	I quite liked Predictor but it was unbelievably unreliable at the beginning and that continues as far as I can tell. __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

4th January 2005, 05:32	#14
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	I'll check with some people who are doing it that I know. Still obviously based on BOINC so the problem of too high a priority may well still be there. __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

9th February 2005, 15:56	#15
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	Any response from them, Mike? __________________

10th February 2005, 03:33	#16
MikeTimbers Messy Yoghurt Eater Join Date: Jul 2001 Location: London Posts: 4,451	Seems that it's not easy to get wus so you have to buy into the whole BOINC thing supporting multiple projects to ensure you have work. The validation approach can mean that you won't get given units to crunch even if there are some unless your particular machine specification matches what P@H want. I'm still staying away... __________________ "There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know."

10th February 2005, 04:13	#17
BennyRop Senior Professional Join Date: Dec 2002 Location: Kodiak, Alaska Posts: 382	I saw the list of this month's bug reports for PaH - and for a project that's supposed to be over 2 years old (they took part in Casp 5) - they are having quite a few major crises.. So the temptation to help out a project that did better than DF in Casp 5 evaporated quickly. __________________

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode