Are only the results with the lowest energy structure kept?

---

The clients report back all their structures (and scores). We permanently keep the reported scores for all workunits, and the full structures for the lowest 5% energy workunits. Actually, over the summer we kept all the structures, but it completely filled our hard drives!

Are only the results with the lowest energy structure kept?

---

Soren, keep in mind that Rhiju is talking about the bakerlab database... not what you will see in your Rosetta@home results page. These are dropped over time. Only the active and recently reported results are maintained there.

It just depends upon whether you are asking about what you should expect to see... or about how the project team uses the science results.

Perhaps your question was more along the lines of whether any of the results OTHER then the lowest energy are "interesting". As Rhiju says, not only do they keep all of the scores, they keep the full structures for the lowest 5%. If they tweek some improvements in to Rosetta, these lowest 5% are probably the ones that will be most successful to apply the new improvements to.

The lowest energy score is not always the "best predicted structure". But it's a good sign you are close to the correct answer.

---

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/

If the structure is already experimentally determined (the "Native" structure), why is it necessary to go through the computations?

---

If the structure is already experimentally determined (the "Native" structure), why is it necessary to go through the computations?

That's a great question. And is the very key point that interests me the most about the project. If we already KNOW how tall Mount Everest is, why measure it? Well, we want to test a new means of measuring and locating things on earth, such as the GPS system. In fact we might calibrate the new methods based on the known figures. This is much like what Rosetta is doing.

Science already KNOWS the structure of SOME proteins. If changes are made to Rosetta which are hoped to bring about better predictions, how will you know if your changes actually were improvements or not if you only run the new program against proteins where noone knows what the native structure looks like?

If our new method alters our prior best model, we have no way to know for sure if the new result is actually an improvement or not. So, they run against known structures to see if they can create a prediction that is indeed an improvement over prior methods. Or perhaps arrive at the same answer as previously, but do so with less computing power.

The ultimate goal is then of course to make consistently good predictions in a fraction of the computing time, and to make improvements and refinements until you feel you have high confidence that the prediction is very accurate. And then you can apply Rosetta to the thousands of proteins of unknown structure, and take steps to produce medicines and vaccines based on your predicted structures, and have the confidence that basing further research and investment in that prediction is prudent.

---

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/