From what I gather, Folding@home will take (very approximately) 100 years to complete its current mission. Does this apply to Rosetta, or will there always be new proteins to simulate?

Also, if we were able to let loose an infinitely fast CPU on the project, I assume we would still need to research the results for a while, and that real practical applications would only result after that.

An infinitely fast CPU would only be useful if you understood how proteins work. Rosetta seeks to discover what to ask such a powerful machine to do. Researching various ideas as to "what makes 'em tick" and how they interact with molecules and proteins around them. So, still much human ingenuity and interaction required.

Bottom line, regardless of speed, noone knows exactly how to tell a supercomputer to solve a protein structure consistently. That's the goal. So yes, real applications of such knowledge follows obtaining the knowledge.

---

Rosetta Moderator: Mod.Sense

That said, CPU time is still a limiting factor because the sample sizes are potentially so vast and the problem so complex. But CPU time is only a part of the jigsaw ;)

---

Okay thanks for both your input - it would seem that human ingenuity is required too. Do you you mean that vast amounts of the human side of the work is needed before all the PCs calculate their stuff (as well as after, which is a given).

In any case, for whatever Rosetta@home is calculating on the countless PCs everywhere in the network, is there an 'end point' or 'goal', or will the (number-crunching side of the) work go on theoretically forever?

Okay thanks for both your input - it would seem that human ingenuity is required too. Do you you mean that vast amounts of the human side of the work is needed before all the PCs calculate their stuff (as well as after, which is a given).

The human side of the work usually goes along the lines of "if I add in factor Y to the Rosetta algorithm then we should get more accurate results". The scientists then draw up a small batch of work units for internal testing. If the internal tests are stable a big batch of work units is produced for the public to crunch and provide a statistically valid sample. We crunch the work units and send the results back to be analysed.

If the scientists see that the results were good then they will decide to add the new factor into the next version of the Rosetta software to be released to the science community.

In any case, for whatever Rosetta@home is calculating on the countless PCs everywhere in the network, is there an 'end point' or 'goal', or will the (number-crunching side of the) work go on theoretically forever?

Rosetta@home's goal is to design and improve a software tool capable of analysing proteins and their reactions to other chemicals. There is no specific goal of "we will analyse 10,000 proteins" so there is also no specific time limit. In theory we could keep improving the software for eternity.

The Rosetta software that we are helping to design here is currently in use by a number of public and private institutions around the world. One public project that uses the Rosetta software is World Community Grid's Human Proteome Folding project.

---

From what I gather, Folding@home will take (very approximately) 100 years to complete its current mission.

If I may ask, what is your source for this information?

---

The human ingenuity comes in to play when an idea on a new approach to the problem comes up. One such example. On the fold.it game they've added a new setting that allows you to define the degree to which clashes are taken in to account as the system does folding.

A human thought "hey, maybe we're taking the clashes too seriously, and could find better energy levels if we provided some means of pushing through clashes" and then from there you have to try it out and see if it helps the predictions or not.

---

Rosetta Moderator: Mod.Sense

I'd say the ultimate goal of Rosetta is devising a program which can predict the correct fold in one attempt. When that goal is reached, distributed computing may not be necessary to predict a correct fold.

But then we have the aim of predicting a correct docking between two proteins in one attempt. Or to design a protein to suit a particular purpose in one attempt.

Both are functions already included within the Rosetta system.

---

I'd say the ultimate goal of Rosetta is devising a program which can predict the correct fold in one attempt. When that goal is reached, distributed computing may not be necessary to predict a correct fold.

But then we have the aim of predicting a correct docking between two proteins in one attempt. Or to design a protein to suit a particular purpose in one attempt.

Both are functions already included within the Rosetta system.

And THEN we have to prove that the above predictions are correct and WILL NOT FAIL under any circumstances. Because if they do fail, even once, then something may go tragically wrong. And in Science that is not a good thing! In short, this may be a VERY long term project!

---