CASP5 results

[Brian the Roman]

Howard;
could you possibly give us a lay-man's rundown on how we did at CASP5? It doesn't have to be exhaustive, just an overview of how close we were overall and if applicable any particular proteins on which we did particularly welll (or poorly for that matter). Notwithstanding the stats discussion going on at present, it's the scienece that motivates me much more than the stats - they're just for some fun and the odd bit of extra motivation.

ms

[Kileran]

agree'd. even if your news isn't great, we'd just like to know something.

[Brian the Fist]

I got some nice graphs I have to put up, I just haven't had the time yet. Will do it ASAP though.

[m0ti]

Thanks for putting the graphs up, Howard!

Actually the results are pretty good. If I remember correctly we only did about 1 billion folds per protein. And the latest algorithm not only gave a speed up (about 2X) but improved results as well. I think that with the new smart algorithm (when it comes out) we could produce much better results.

Just out of interest, but instead of continuing to fold against the proteins from CASP 4, why don't we do the proteins from CASP 5? I'm sure we'll be able to produce great results and it'll be interesting to see how much better the new algorithms' results are.

[Brian the Fist]

I assume you are referring to our 'phase II' on the Science part of the web site? Anything there is definitely written a long time ago and now subject to revision. In short, we will definitely NOT be looking at CASP4 as that is old news.