Hi,

what does "low Energy" mean in the visual window at rosetta.

Tahnks

Daniel

It's not clear exactly what you're asking. Rosetta is looking for the best and most natural shape for the protein. The "lowest energy" shape is another way of saying best. In the screensaver, I think the "low energy" frame shows the best shape that your computer has found so far.

Neal

The energy that they're referring to is the potential energy of the configuration. Individual amino acids in the chain that make up the protein exert some force on one another, and there is a potential energy function associated with these forces. For example, if two links in the chain repel one another, then pushing them together requires work (or energy) to be put into the system, since the potential energy grows larger as they get closer; on the other hand, two links that attract one another will have a lower contribution to the potential when they are closer together, and pulling them apart requires work to be put into the system. As nneal has already pointed out, the lowest energy shape is the most natural shape for the protein. One way to think about this is to consider two weights connected by a spring. There is some natural length to this spring, and the potential energy of the spring at this length is a minimum. If I push or pull on the spring, I change its potential energy, which generates a force in the opposite direction, pushing it back towards the lowest energy state. Proteins are kind of like a bunch of little weights connected by springs of different stiffness.

So what are the value ranges?
eg. >50 is not really good and therefore unlikely to be the true structure while values < -500 might be approaching the true form?

Also I'm assuming the RMSD needs to approach 1 to validate the likely hood that the predicted low energy form closely matches the native form. Is that even remotely correct?


I guess the new community, me included, want to know if they are making a difference and getting good results from their hardware.

There is a daily lowest predictor but it does not say what that prediction was in values. If there was a top 100 list of lowest predictors with their scores it would give some idea of things moving in the right direction.

Not sure if I'm even making sense to myself here.

Hello Jacques, welcome to Rosetta@home! Thank you for helping out with your machine.

Some of what R@h does is pure research. And often they are testing a new method of making a prediction. In tasks where you see an RMSD, it means that the structure being run is already known. So your machine is running the new method, and seeing how the structure that it predicts compares to the known structure. In tasks where you do not see an RMSD, it means the structure is unknown, so the lower the energy level, the more likely the predicted structure is close to the natural, or native structure. IIRC, an RMSD of 1.0 would mean your model exactly matches the known native structure.

---

Rosetta Moderator: Mod.Sense

Thanks Mod.Sense for clarifying.

I guess it is pure curiosity to understand some of the values being shown while I was looking at one cycle passing by.

I'm guessing what I am seeing as a value of about RMSD 20 then means it is about 400 times deviated from the native form. Not sure what would quantify as a deviation either.

The lowest energy I saw was about -500, but I have no concept of scale to even fathom if that is a good value or simply run of the mill.

Assuming -500 is a good score, I then deduce that the model could potentially fold into that shape but it would be vastly different than the known form.

Am I right in thinking if it has a very low energy and RMSD close to 1 it would look very similar to the known form and quite easily take or fold into that shape?

Assuming -500 is a good score, I then deduce that the model could potentially fold into that shape but it would be vastly different than the known form.

That puts it very well. I think the raw energy scores are essentially only limited by the number of amino acids in the protein. More AAs brings more potential bonds. If you look at the project homepage right now, there is discussion of proteins that have more than one native conformation ("shape-shifting proteins"). The RSMD can only compare to one or the other, so what you describe is probably exactly how a shape-shifting protein would look when compared to the other conformation.

Am I right in thinking if it has a very low energy and RMSD close to 1 it would look very similar to the known form and quite easily take or fold into that shape?

Yep! You've got it!

Since you are essentially trying to compare a complex 3 dimensional chain of things to another chain of things, you can have a lot of things right and still not visually appear the same. The RSMD sort of gives you a score of the fit that is not deceived by the eye of the beholder.

Thank you for crunching!

---

Rosetta Moderator: Mod.Sense

Thanks Mod.Sense!!!