With the new methodologies being developed, will there be a point at which we go beyond the needle-in-the-haystack decoys and start clustering around the actual structures of protiens?

And is this the goal before (from what I understand) the project moves into the design/docking phase?

---

Keith, since the science team might have a hard time reading every message during the CASP7 period, you might want to ask again in Discussion on Dr baker's journal or some other appropriate topic in Science subforum.

I've asked this question in the past myself.

Obviously a lot depends on the protein topology itself. Some are particularly hard (e.g. 1tul), which is why the RMSD-Energy chart distribution of red dots is so different from one protein to another.

AFAIK the energy function is quite accurate, but the problem is sampling the conformational space.

---

Best UFO Resources
Wikipedia R@h
How-To: Join Distributed Computing projects that benefit humanity

Thank you.

I wasn't sure which board to ask this as it relates to both crunching and the science.

I imagine that CASP7 targets are falling from the sky. CASP must have released a large bag this round.

---

Over time, as more is learned about how the theorectical models compare to the physical reality, the modelling will get better and better. So, as Dimitris says, at present the RESULT produced is quite accurate in general (we'll watch the CASP results for confirmation of this), but, as you pointed out, the present methods require 100s of thousands of models crunched to arrive at that pretty accurate solution.

So, over time (I'm talking years) you should expect to see both a more consistently accurate model (rather than hit and miss on one protein verses another), and to see this model produced with less computing power required for the task.

This is exactly why the project team has been putting out new releases, and trying new techniques over the past 6 months or so (less now with CASP, keep it stable is the goal for now). As they try out these different ideas, they will strike on a few that reinvent the state-of-the-art.

I took an Artificial Intelligence class in college. The professor defined AI as being anything that is considered significantly difficult to do on a computer. And that once it becomes commonplace, then it really isn't considered AI anymore. Playing chess might be a good example. Used to be cutting edge stuff for a computer to beat a chess master. But today, I doubt anyone would consider a computer playing chess to be exhibiting intelligence.

I think it will be much the same with protenomics. I think that today these problems are cutting edge stuff for a computer to find a great model. But in 10 years or so, it will be commonplace. And even if Rosetta's algorythms didn't change a bit during that time, having computers that are 100 times faster than today will mean you'll be able to crunch 100,000 models on a couple of dozen "home" computers, rather than today's thousands of PCs.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

Thanks Feet1st.

I could barely afford one of today's computers.

As for AI;

Greetings Professor Falken.

---

As for AI;

Greetings Professor Falken.

At the risk of drifting too far off topic, there is a certain symmetry here. The teenage computer whiz in that movie lived in Seattle. :)

---

That is so wierd. I just bought a Slurpie from Seven-11

---