1 reply [Last post]
brow42's picture
User offline. Last seen 8 weeks 22 hours ago. Offline
Joined: 09/19/2011
Groups: None

Could you describe more about how rosetta handles hbonds, how they are scored, and how the Hnet is scored? Specifically, I was looking at # of bond and directionality of bonds.

Here is a picture of 10 hbonds to the same sidechain: http://fold.it/portal/files/chatimg/irc_269484_1445571359.png

In the blog post, it says acceptors count as fully satisfied if they have at least 1 hbond.

So, specific questions, but please also elaborate if I didn't hit the right question:

1) Are acceptors purely electrostatic (no lone pair directionality)

2) Is there a big score jump if an hbond matches a donor directionality condition? An acceptor? Two bonus for both, or no bonus unless both?

3) is the preferred direction for donors along and opposite the X-H bond, X = N,O?

4) Is this partial covalent bond the difference between a red network hbond and a blue network hbond? What is the difference?

5) how did I get 10 hbonds anyway?

sboyken's picture
User offline. Last seen 3 years 27 weeks ago. Offline
Joined: 09/01/2011
Groups: None
In FoldIt, h-bonds are pairwise and calculated independently

@brow42, your questions are really good -- I'll answer them each below (I'm the scientist in the Baker lab who is working on the method for hbond networks).

1) Are acceptors purely electrostatic (no lone pair directionality)
No, the hydrogen bond energy is dependent on the geometry (distance and angle) between the donor hydrogen and the lone pair electrons of the acceptor. There is also a general electrostatic attraction between the partial positive charge of the donor and the partial negative charge of the acceptor. In FoldIt and Rosetta, the electrostatic energy and the hydrogen bond energy are calculated independently (even though in reality these forces are heavily dependent). For the networks, we are concerned more with good hydrogen bond geometry than electrostatic attraction.

2) Is there a big score jump if an hbond matches a donor directionality condition? An acceptor? Two bonus for both, or no bonus unless both?
For an individual hydrogen bond, more-so both -- but it will not big a "big" jump -- the bonus is for the network. The better the geometry between donor and acceptor, the better the score for each h-bond. A hydrogen bond has to have a good enough geometry (score) to be included in the network, and the important properties for protein design are related to the connectedness of the network -- multiple hydrogen bonds that are coordinated in favorable ways, which is what we are trying to capture.

3) is the preferred direction for donors along and opposite the X-H bond, X = N,O?
It depends on the geometry between the donor hydrogen atom and the lone pair electrons of the acceptor atom; different acceptors have different numbers and orientations of lone pair electrons ("hybridization"). Rather than directionality, think geometry. In many cases, the "best" geometry is very similar to what I think you are saying, which is more or less 180 degrees (straight-on) between the donor hydrogen and the acceptor.

4) Is this partial covalent bond the difference between a red network hbond and a blue network hbond? What is the difference?
Yes -- kind of. Again, it's all about the geometry. The red indicates an hbond with worse geometry, so by improving geometry, it can become blue (a better hydrogen bond) and included in the network.

5) how did I get 10 hbonds anyway?
in order for things to be calculated efficiently in FoldIt and Rosetta, we have to do things pairwise, which means if all 10 possibilities exist, we score all 10. All of these 10 hbonds that you're seeing are possible, but in reality, each acceptor (of this type) can only accept 2 hydrogens and all 10 cannot exist simultaneously. Some of these hydrogen bonds are not as good as others (worse geometry and hence worse score), so not all of them would be allowed in a network. Short version is you stumbled upon a kind of weird case. Usually, this many hydrogen bonds to a single acceptor would not be possible simply because of the backbone restrictions and things clashing with each other.

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