Unofficial "new protein size" thread

[Anteraan]

Let the guessing begin again. For some reason, the number 94 is in my head.

[pointwood]

2

seriously, my be is 105

[Grumpy]

112...well, it is better than 129

[PCZ]

Prepare for an invasion of Cows.

I guess 58

[tpdooley]

154, and an extremely tangled 154 that runs even slower than you'd estimate.. (If I'm gonna bet wrong, I'll make it so I'm glad I'm wrong.

[iggy]

58 - and I hope that this time my prediction counts!

[PCZ]

iggy

No you cheated same as me

[gopher_yarrowzoo]

if it is 58 then we are sunk unless we get everyone onboard...
Hmm I predict 81

[Gortok]

I'll guess 108, again....:sleepy:

If not ( and I HOPE not...) I'd say 58

G

[iggy]

Originally posted by PCZ
iggy

No you cheated same as me

Running daemon is too easy!

[Anteraan]

So, running the daemon before the new protein is introduced (technically) allows you to see the new protein size? So this is unofficial confirmation of 58AA?

</me runs to local store to get new mobo/CPU upgrade>

[Grumpy]

If it really is 58 and not a leg pulling excercise, Ars are so going down to #3

[2fast4u]

edit

[Anteraan]

Yea, how does that work? And people were afraid of "stats cheating" by allowing full credit for uploads from 24-48 hours...? At least everyone had that option (or do we?).

He is a mirror, that much I know. Maybe hijinx like this were one of the reasons that mirrors weren't used previously. Either way, the ethics of this is a bit disappointing to me.

[Grumpy]

I think there is a leak from several sources This was predicted, no system is foolproof, there are too many fools trying to break it

[rel279]

I'm sry for this post 105 it is all ready posted

[Pascal]

146 - don't know, why.

[jonnyw]

I'm not even gonna attempt a guess this time.

First I was accused of cheating, and then my prize for getting close

as posted by grumpy:

An A3 print of me in the shower Folding

[Robert Radford]

Forgive the newbie but what are you guys talking about?

As I live in Toronto, know of Mount Sinai hospital and think I understand the reasoning behind the science (I did stats at university), I have downloaded the software and cranked out a number of work units but the point and ranking system is obscure and the FAQ even more so.

Now you have a forum topic about a number but don't mention what it represents. Is it the average age of participants or the size of a target protein? Is it the protein we are now doing and what is it 48 of? Or are we not doing a new protein as I don't see any change in my ranking other than the number of work units I've submitted?

Label me confused but hopeful the universe will eventually make sense.

[Brian the Fist]

Originally posted by Anteraan
Yea, how does that work? And people were afraid of "stats cheating" by allowing full credit for uploads from 24-48 hours...? At least everyone had that option (or do we?).

He is a mirror, that much I know. Maybe hijinx like this were one of the reasons that mirrors weren't used previously. Either way, the ethics of this is a bit disappointing to me.

Just to clarify, anyone can download the 'daemon' from the download page, and get updates 1 day prior to release. (See the documentation included with the download for details).

[Anteraan]

Thanks for the clarification. I certainly wasn't aware of that (obviously, so I take back my previous comments), but I'll take advantage of it in the future. Perhaps that could help the servers some at update time.

The number we speak of is the number of amino acids in the protein being worked upon. The new one has 58, while the previous one has 129.

The scoring system works as follows:
Points are awarded based upon the generation number that the client is working on, and the formula is:
Points = INT{100*[SQR(gen #)]}

with the exception of gen 0, which awards 100 points.

Tying this all together, the new, smaller protein will be quicker to fold, and thus points can be earned more quickly, which can affect team and individual rankings.

[Robert Radford]

Thanks for the info on the point system.

I was a little bemused at the start when my numerical (higher is worse) ranking was higher than the number of active participants. I guess that there is a residue in the statistics of people who once were active but are no longer so.

As to the latter, I see that the number of active people is less than 10% of the total number of people registered. Is this cause for concern?

[AMD_is_logical]

Originally posted by Brian the Fist
Just to clarify, anyone can download the 'daemon' from the download page, and get updates 1 day prior to release. (See the documentation included with the download for details).

The files that the daemon downloaded (gcc linux) were *.tar.gz , but gunzip couldn't handle them at first. I finally figured out that there was an extra 175 bytes that had to be lopped off the front (the digital sig?) after which I had no problem.

That let me upload all old work and start crunching the new protein on my cluster (offline, of course) before the DF server changed to the new protein. (The DF server can only collect data for one protein at a time, so any old work uploaded after that change is discarded, and counts only for stats.)

[tpdooley]

Originally posted by Robert Radford
I was a little bemused at the start when my numerical (higher is worse) ranking was higher than the number of active participants. I guess that there is a residue in the statistics of people who once were active but are no longer so.

As to the latter, I see that the number of active people is less than 10% of the total number of people registered. Is this cause for concern?

To be considered "active" you have to upload at least one structure for the current protein. Some of those that are no longer active moved to other projects, some have had all their folders taken away, and others have had their clients crash, and just assumed that they're still running, or they've gone on vacation and left the farm folding offline or just left it off, or they've left the project until the client ends up as reliable as it was at the end of Phase I. There should be lots of inactive people with scores that represent their contribution during Phase II.

For the last 1.75 years that I've been with this project, the number of active users has always seemed to hover around the 10% mark of the total number of registered users. It would be nice to see more of the registered users active so we can blaze through some of these slow proteins.. and get to test out new approaches.

[Scotttheking]

Originally posted by AMD_is_logical
The files that the daemon downloaded (gcc linux) were *.tar.gz , but gunzip couldn't handle them at first. I finally figured out that there was an extra 175 bytes that had to be lopped off the front (the digital sig?) after which I had no problem.

That let me upload all old work and start crunching the new protein on my cluster (offline, of course) before the DF server changed to the new protein. (The DF server can only collect data for one protein at a time, so any old work uploaded after that change is discarded, and counts only for stats.)

Oh how nice it would be if there was an automated way to do that.
Imagine, no production wasted due to the server having already switched.
I'd be happy.

[Mumbles376]

So what determins the size of the next protein is it done randomly from a list so that you get a mix of large and small ones to crunch through or is there a more scientific aproch to when they want to crunch each one ?

Forgive me I am only a N00b here and stilll finding my feet 3 days into crunching now so got a long way to be like the rest of you

[Brian the Fist]

We have been going through protein that were previously folded in phase I to compare how they fare in phase II. The order is somewhat arbitrary and depends on various factors, we try to vary the protein size, shape etc.