Preliminary CASP8 results

Message boards : POEM@HOME : Preliminary CASP8 results

To post messages, you must log in.

AuthorMessage
Timo Strunk
Volunteer moderator
Project administrator

Send message
Joined: 6 Sep 07
Posts: 560
Credit: 10,729,563
RAC: 0
Message 2823 - Posted: 2 Sep 2008, 11:27:15 UTC

Dear Participants,

we used the recently released slew of experimental structures for the CASP targets to compare with our results. Below you find a preliminary analysis, comparing the standard TM-Score with the best structure of all serves with our two contributions POEM-QA and POEM. According to this initial analysis we are very satisfied with the POEM-QA results, which succeeded to select a good server prediction in almost all cases. The POEM group did also quite well in the Template Based Modeling sector, lacking a bit in the free modeling sector.
Our analysis continues and we will keep you posted on the details.

Thank you for your tremendous support, which made this possible !

Preliminary graphs comparing the best of the submitted server structures with the best of our structures can be found
here (Targets 389 - 434),
here (Targets 437 - 474) and
here (Targets 476 - 493).

A TM-Score of one indicates perfect agreement, while zero is total failure. A TM-Score above 0.16 is considered a non-random prediction,
due to experimental uncertainties TM-Scores of 0.9 are hardly ever reached in practice and can be considered perfect predictions.

Best regards,
the POEM@HOME team.
ID: 2823 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
J Langley

Send message
Joined: 8 Mar 08
Posts: 13
Credit: 64,435
RAC: 0
Message 2826 - Posted: 2 Sep 2008, 19:01:37 UTC - in response to Message 2823.  

Thanks for the update - hearing about the work our computers are doing is one of the main motivations behind which projects I choose to support.
ID: 2826 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
P . P . L .
Avatar

Send message
Joined: 10 Feb 08
Posts: 78
Credit: 1,007,923
RAC: 0
Message 2827 - Posted: 3 Sep 2008, 3:11:18 UTC

G\'day Timo.

Very good news, but can i ask what is the difference between

POEM and POEM-QA i had a look around and couldn\'t find anything.

thanks.

pete

ID: 2827 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ananas

Send message
Joined: 17 Oct 07
Posts: 241
Credit: 2,500,485
RAC: 0
Message 2828 - Posted: 3 Sep 2008, 8:00:27 UTC
Last modified: 3 Sep 2008, 8:06:36 UTC

QA stands for \"Quality Assessment\" (QS / Qualitaetssicherung in german I think)

You find some more on this page.
ID: 2828 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
P . P . L .
Avatar

Send message
Joined: 10 Feb 08
Posts: 78
Credit: 1,007,923
RAC: 0
Message 2975 - Posted: 7 Dec 2008, 4:05:44 UTC

Hi.

Is there any news on how POEM went i think some results are out.

thanks.

pete.

ID: 2975 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Timo Strunk
Volunteer moderator
Project administrator

Send message
Joined: 6 Sep 07
Posts: 560
Credit: 10,729,563
RAC: 0
Message 2976 - Posted: 8 Dec 2008, 0:06:36 UTC

Hi everyone,

I just arrived home from the CASP meeting, where the results were published (so we saw them ourselves just now, also).

We are very happy with what we could achieve and there are some especially interesting targets where we did very good and some even more interesting, where we didn't.

As it is already quite late over here, I will prepare a bit longer statement tomorrow at work.

Bye,
Timo
ID: 2976 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Timo Strunk
Volunteer moderator
Project administrator

Send message
Joined: 6 Sep 07
Posts: 560
Credit: 10,729,563
RAC: 0
Message 2977 - Posted: 8 Dec 2008, 16:12:10 UTC - in response to Message 2976.  

Hi again,

So first of all two targets, which we did good and bad at:

One target we are quite proud of is T0423
If you look at all the submitted models of this target, we come out first:
T0423
As POEMQA did refinement of the input structure, this means, that our way to refine really worked, as the structure is better than any of the ones we used as input.

One of the targets, which I found really interesting was target T0401:
T0401. POEMQA submitted a bad model (GDT-TS 27/100) and a good model (GDT-TS 65/100).
As the simulated energy of the bad model was quite a bit 'better' for the bad model, we investigated why this happened, as afterall there were structures of good topology in the populations we rated. It occurs, that T0401 naturally occurs as a dimer, so two of the same proteins are adjacent to each other; we however started the simulation from a monomer, because it was not known, that the protein occurs as a dimer before CASP.
The interesting part here now is, that if the protein would be isolated, our conformation could really be lower in energy, the interaction energy between the two domains however stabilizes the structure.

As I already said, overall the participation of POEM and POEMQA was a success. POEMQAs results were better than those of POEM, because most of the time POEMQA had access to populations of starting structures, which were already nearer the native structure.

Bye,
Timo!
ID: 2977 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
J Langley

Send message
Joined: 8 Mar 08
Posts: 13
Credit: 64,435
RAC: 0
Message 2980 - Posted: 8 Dec 2008, 18:43:35 UTC - in response to Message 2977.  

How did POEM do compared with other groups (such as Rosetta)?

And how did other groups' resources compare with POEM's? (Do POEM's models work better per FLOP?)
ID: 2980 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Steven Pletsch
Avatar

Send message
Joined: 18 Nov 07
Posts: 84
Credit: 1,282,872
RAC: 0
Message 2988 - Posted: 23 Dec 2008, 13:15:55 UTC - in response to Message 2977.  

It occurs, that T0401 naturally occurs as a dimer, so two of the same proteins are adjacent to each other; we however started the simulation from a monomer, because it was not known, that the protein occurs as a dimer before CASP.
The interesting part here now is, that if the protein would be isolated, our conformation could really be lower in energy, the interaction energy between the two domains however stabilizes the structure.


I am curious, is there any signs when looking at a model that would indicate which way a structure should be approached other than a "guess" ? It seems a bit unfair that such a structure would be fed blindly to be folded if there were no means of discernment. If there is a means, what specifically would you look for? Are there certain amino acid sequences that commonly occur, or a general shape that would indicate the nature of the protein?

Thanks
"Every passing hour brings the Solar System forty-three thousand miles closer to Globular Cluster M13 in Hercules -- and still there are some misfits who insist that there is no such thing as progress." - Kurt Vonnegut
ID: 2988 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : POEM@HOME : Preliminary CASP8 results


Copyright © 2017 KIT-INT