10 replies [Last post]
adonai's picture
User offline. Last seen 49 years 35 weeks ago. Offline
Joined: 06/08/2008
Groups: None

The concept of the high score has to be based upon the designers approval of a known "good" protein design submission.

How do you reward "good" protein design if the desired protein structure is an unknown?

Aren't you potentially limiting user creativity based on precedent?

Thanks

Jwb52z's picture
User offline. Last seen 30 weeks 1 hour ago. Offline
Joined: 05/11/2008
Groups: None
That's not what it is based on at all.

It's based upon a mathematical formula that will test to see how much energy is needed to hold the protein in its current state that you have put it in at the time. The lower the energy required, the higher your score becomes.

adonai's picture
User offline. Last seen 49 years 35 weeks ago. Offline
Joined: 06/08/2008
Groups: None
Thanks for the prompt reply.

Thanks for the prompt reply.

Thats what I'm trying to highlight, the formula details "good" design based on what we already understand about protein.

So based on your reply, good design is energy efficent, doesn't this seem to exclude a whole multitude of protein design that still is potentially effective, yet energy inefficent?

I guess I'm having trouble trying to reconcile what appears to be a paradox.

Course I'm just a noob!

Jwb52z's picture
User offline. Last seen 30 weeks 1 hour ago. Offline
Joined: 05/11/2008
Groups: None
I'm guessing biology isn't your strong suit

Proteins wouldn't fold in a bad way to use more energy than absolutely necessary in nature. That's part of the reason they fold, that and they have to fold to work at all. It doesn't work like you think it does.

adonai's picture
User offline. Last seen 49 years 35 weeks ago. Offline
Joined: 06/08/2008
Groups: None
I guess civility isn't your

I guess civility isn't your strong suit?

I'm not interested in examples in nature, these protein designs are what I consider known. Foldit "about"...

http://fold.it/portal/info/science

...details protein design as an upcoming feature and talks at length about users creating "artifical" proteins.

So my question is still how do you reconcile the math formula which rewards known protein design while trying to achieve new protein designs?

Lastly if you don't have anything constructive to add, just troll up another area of the forum, I'm sure your vast expertise in Biology as well as the sizeable chip will be so useful elsewhere.

Jwb52z's picture
User offline. Last seen 30 weeks 1 hour ago. Offline
Joined: 05/11/2008
Groups: None
It's just like any other formula

You plug in the data about the different variables, in this case the parts of the protein, and then you evaluate it and solve the equation or set of equations.

Joined: 06/09/2008
Groups: None
It's like designing a brick wall

It's like designing a brick wall.

You seem to think that brick walls are designed by comparison to previous brick walls. Well, in the real world, maybe, but that restricts the creativity of architects.

Fold it instead just knows about bricks, mortar, and gravity. We think we know enough about bricks to calculate what the wall will be like once it's built.

If you build it and the brick-calculations say it will fall, I don't want to be there. Go ahead, be creative, but the physics is known.

Does that help?

Joined: 06/10/2008
Groups: None
I wondered about that too

I wondered about that too and I think I got it now.
Thanks
Ben

Joined: 05/27/2008
Groups: Minions of TWIS
Good question

Actually, Adonai brings up a very good point. Although we are not trying to conform our proteins to a designer's preconceived idea of what they should look like, the best design can only be as good as the energy function we are trying to optimize. Even a small change in how the energy function is calculated could have a huge impact on the optimal structure. However, my impression is that this part of structure prediction is much farther along than algorithms that optimize energy - that's why we are manually folding proteins and observing how their energy changes instead of perturbing the energy function and letting the computer fold them.

Also, contrary to the earlier comment, nature does create proteins that are not "optimal" in terms of energy - most active sites are essentially large voids.

Joined: 06/25/2008
Groups: None
The answer

The score is not calculated by comparing it to known structures. Unnatural (synthetic) proteins can often adopt severl very different structures but in nature, the usefulness of a protein (its function) is determined by its shape. Therefore, evolution has selected proteins which only fold into one functional shape or a number of similar shapes. This has been shown by unfolding natural proteins in solution (with chemicals or by heating it up) and then letting them fold again - they adopt the same structure.

When a protein folds, it releases energy. To subsequenly unfold it requires putting the same amount of heat back in. The most stable structure, and hence the answer to the folding problem, is the structure that releases the most energy. Atoms are 'sticky' and packing them together releases energy. Similarly, some sidechains are charged and energy is relesed when opposite charges come together. Conversely, you need to put energy into the system to bring like charges together, or to push atoms too close together. As it is possible to calculate how close together atoms are, and how close together charges are, it is possible to calculate the amound of energy released upon folding.

As for the hydrophobic/polar side chains; Hydrophobic side chains ('oily') need energy input to mix with water (you have to shake a vinegrette to mix the oil and water). Therefore its best to have them together in the interior. Polar sidechains, on the otherhand, release energy when in contact with water and so are best on the exterior.

In actuality, the formulas are a bit more complex that I suggest here, as there are changes in energy associated with changing side chain torsion angles and a special term for hydrogen bonds.

Joined: 07/03/2008
Groups: None
LPLR: the widespread Law of Path of Least Resistance.

I can't vouch for the details, but the idea of rewarding low free energy states complies perfectly with the Least resistance rule of thumb that all objects, both living and non living follow.

I prefer the meandering river analogy to explain it. A river on its journey to sea will meander because it encounters less resistance by looping around rocky structures (say). Meanders are hardly the "shortest" paths, if fact they tend to be the longest! But the quantity being minimised is not distance, but gravitational energy, so big loops make sense if the land provides less resistance.

This is the way I, at least, have reconciled the idea that proteins, will look for a low energy state in their folds.

Hope that helps.

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons