Genome listing, how to read it?

[Jorden]

I think this is a question for Martin, but maybe someone else might give me an insight as well.

Looking at Genome's Matroxusers's list I'll take myself as the example.

First in the list below is my name, then my rank, followed by the Units, followed by my Delta Units.
Jorden 19 56 26

Okay, now the question:
I've downloaded and crunched a unit of 92 genes, one of 84 genes, one of 68 genes and am currently doing 92 genes again. All of them are crunched in 30 iterations.

Now, if I sent back the 92, 84 and 68 gene units already, how come I am depicted as having gotten (and sent back) 56 (gene?) units, with a delta of 26?

It's probably uncomparable to Seti, so I wonder how the calculations for this thing go.

Anyone who can shine a light on that? I'll add it to the FAQs then

Jord.

P.S: Warning about getting Genome Units. I got two 99 gene units in which crashed my genome client. If you see the filter starting 99 units, or your client isn't running as it should, not even when you try to restart it, then just delete the rotamers.lib and the input.pdb ... then upon starting ghclient again, it should download a new unit

[This message has been edited by Jorden (edited 08 March 2001).]

[Liquid Snake]

Well, the numbers you see at the beginning are actually the amino-acid length. Higher is obviously longer, and they take more time to process, but you get paid more for your effort. Longer aa-length = more WU credits. So actually, when you download one data block, it's one gene. I think, when you crunch a 100 aa-length unit, you can get 30 units of credit. Works different from SETI, which is why you see numbers such as 121.14 units on Genome's stats page.

Oh, and I'm probably completely wrong, so feel free to correct me

[Jorden]

LS, can you explain that again, in real English?

Besides, as I stated above, every unit I had (and have) was crunched in 30 iterations already. The only thing I found different were the Last Iteration Matches in Genome Voyeur, being added up through the same unit.

But even they don't add up to my units on Martin's page

Jord.

[-=Narcissus=-]

Ok, let me have a shot at it.

First of all, every wu (I'll call it block from now on) takes 30 iterations, or intermediate steps. That's just how the program works.

Now the 92, 84, 68 units you saw per block were the animo-acid length. It's just a property of a given gene. Some genes are small, and some genes are BIG. Better ones take longer to process.

You have actually returned 3 genes or blocks, as is written on the official g@h stats page. That part you got right. You just don't add up animo-acid length to get your total points.

Mixed up? Good. Now it gets better!
The guys at g@h were conscious that some blocks took longer to compute than others. So they made up a point system. A 100 animo-acid block will give you 30 points. Shorter proteins take less time to calculate and thus give back less points (I don't think it's linear).

The stats page gives you the total number of points, as is calculated by the g@h system. That's why adding you aa length doesn't add up.

Got it now?
Fred

[Liquid Snake]

Well, that's better than my explanation.

[Jorden]

Thanks Fred, I got it.

I just wonder as an extra how we can find out how much points we get every time we send a block in

Jord.

[-=Narcissus=-]

Well, if it was linear, you'd have 73.2 points right now. You now have 76.68 points.

So just take the 5% error and learn to live with it.
Take your animo-acid length and multiply by 0.3. That's it.

Fred

[Jorden]

Thanks... added to the FAQs already

Besides, it's another race to freedom, and not one run by "I have 45 genes, yet he has only 10 and is above me in the list".

I now understand

Jord.

[Rattledagger]

Since a 100aa-length is 30 credited, the math should be 0,3*aa-length, but this isn't correct. My last unit was 99 aa, but I got credited 30,25.
If it really was 92aa for Jorden, we get:
30,25 is 99 aa
29,94 is 92 aa
25,97 is 84 aa
20,77 is 68 aa

The 92aa is either a typo, or they're using some weighting due to how complex a gene is. If the latter, it's impossible to calculate based on the aa-length alone. The 92aa is 18% higher credited than expected.

[This message has been edited by Rattledagger (edited 08 March 2001).]

[Mark F]

Yeah, I had one ...ahhhm... gene ... ahhhm... what ever one download is; it crunced fine for 24 iterations and then crashed the client. On a restart, even after a reboot, it would just crash. Had to through away all that work. So, I feel your pain. ARRRRG.

Mark F.

------------------
OH NO, my retractable cup holder swallowed a DVD...
and burped out a movie