Identify unsatisfied polars in H-bonds

Case number:845833-2012329
Topic:Game: Display
Opened by:BootsMcGraw
Status:Open
Type:Suggestion
Opened on:Monday, November 15, 2021 - 05:40
Last modified:Wednesday, April 13, 2022 - 23:30

I've got a beautiful network in puzzle 2070: twelve interface bonds. But I'm getting only 2100 points for it, because only 87.5% of the polar atoms are satisfied.

Honestly, I can't see where I have any unsatisfied polars in my network. I ask that the game dev's please implement a function that shows unsatisfied polars in an H-bond network, much like we can see BUNS by checking a box in the Objectives. I can't fix what I can't see.

(Mon, 11/15/2021 - 05:40  |  15 comments)


bkoep's picture
User offline. Last seen 8 hours 10 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

Yes, that's a good point! We should be able to highlight unsatisfied atoms.

Can you also share this solution so I can make sure there's not a bug in the existing behavior? BUNS atoms are tricky to count...

Joined: 12/06/2008
Groups: Contenders

Solution shared with scientists "82314-2070".

I have an asparagine at residue 16 that has a red bond from the oxygen to the network, so it is "scoring too low to be part of the network". Its nitrogen is bonded to nothing. Could this be the cause?

Funny, removing ASN 16 drops my network score from 2100 to 1800, so I fail to see how it's not part of the network. Conflicting observations.

HuubR's picture
User offline. Last seen 9 hours 45 min ago. Offline
Joined: 04/13/2020
Groups: Gargleblasters

Just want to share my experience with red bonds and unsatisfied networks.

  • A red bond is not part of the HBN, but it does satisfy the polar atom on the residue that is part of the network. I think that could explain your drop from 2100 to 1800 when you remove the ASN and the red bond.
  • In case one of the blue bonds connects to the backbone, it seems that the sidechain of that residue, if it is hydrophilic, also needs to be fully satisfied (see note). In such a case, you can improve the score by replacing that residue by a hydrophobic one: then its sidechain no longer has polars that are counted as unsatisfied.

Note: I don't see the logic why the sidechain should be satisfied when the network only connects to the backbone. I would appreciate it when someone could explain that to me.

Joined: 03/28/2020
Groups: Go Science

Yes, this topic puzzled and sometimes puzzles me still as well. In my opinion this comes from the issue that a bipolar sidechain (serine, threonine or tyrosine) is used in your network. If you take a buried serine for example, it needs to make a bond from it's donor to an acceptor on the other end. But in addition, to the acceptor-part of the serine's atom the donor of another sidechain must be connected in order for the serine to count as fully-satisfied. The irritating thing is that the serine will not glow "red" (indicating an acceptor-bun) when it's donor has an hbond. Foldit-internal it will count as not fully satisfied though and will lead to decreased bonus.
In my opinion the hbnet scorefunction should be corrected so that any cross-Interface-core-bond that has a fully satisfied sidechain on each end of the bond (no matter how else the network looks like) will count as fully satisfied bond. 3 of those should give you full hbnet-bonus. As it is now, you will get decreased bonus if such a residue is only connected with it's donor anywhere to the net which can be frustrating. I try not to use these altogether due to this issue. I typically use a chain of 2 tryptophans (at the ends), 1 aspartate/glutamate and 1 histidine. These can in most cases be made to form a cross-interface chain which is fully satisfied and gives full bonus.

Joined: 03/28/2020
Groups: Go Science

On second thought, another nice way to fix this would be to change the hbnet-bonus-scorefunction so that serine, threonine and tyrosine would count as fully satisfied even if only their donor-part would have an hbond. Like this, there would be 4 (serine, threonine, tyrosine and Tryotophan) AAs which could be used as easy, single-polar terminators for open acceptor-atoms in a net (I would really have loved nature to give us something equivalent as only-single-acceptor but that would have been too easy ;o) ). Currently only tryptophan can do this and it is pretty bulky and therefore sometimes hard to place. This would probably greatly simplify the creation of deeply-buried, fully-satisfied nets and maybe increase their quality.
The resulting question that I have with this goes to scientists: are unsatisfied acceptors in buried serine/threonine/tyrosine an issue? Since they are currently not showing as buns I guess that they are not, so it would be great if the hbnet-bonus-scorefunction could be modified like proposed here.

bkoep's picture
User offline. Last seen 8 hours 10 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

This feature is included in the latest devprev update.

Now, the display for the HBond Network Objective will highlight unsatisfied polar atoms that are incurring penalties.

Joined: 12/06/2008
Groups: Contenders

[and the peasants rejoiced!]

HUZZAH!

HuubR's picture
User offline. Last seen 9 hours 45 min ago. Offline
Joined: 04/13/2020
Groups: Gargleblasters

Thank you, this is helpful, but...
see Feedback Item red spots with blue HBNs

Joined: 03/28/2020
Groups: Go Science

Thanks for implementing the marking of unsatisfied polars in hbnets. This shows indeed that a serine or threonine connected ONLY with its donor to the net shows up with an unsatisfied acceptor which explains to me the bonus-decrease in such a case.
But I am still puzzled by the score-function in hbnets. I find it wrong that the score-function will ALWAYS give a penalty for a single net when there is any unsatisfied anywhere in the net even though it is not buried. I mean that these non-buried unsatisfieds do not pose any problem in my opinion (they will probably not lead to misfolding or so) and they should not be penalized in the hbnet-bonus!
This would allow for more easy termination of nets at the end (as long as the ends are "outside" the core/interface) because then even an unburied lysine or arginine or so could be used as terminator even though not all of its donors would be satisfied.
The criterion for full bonus for a single net should in my opinion be that in the net exist at least 3 fully satisfied (referring to the core-redidues AT THEIR ENDS only) cross-interface-bonds. It should not matter if there are more open atoms in the net as long as these are not buried. Only open BURIED usatisfieds should incur hbnet-bonus penalty.
As it is now I stick almost only to the same net-structure (using tryptophan to fully terminate any open acceptor in the net; see my share on 2093 for an example). Such a net is always 100% satisfied. But with the proposed mod in the score-function, much more diverse nets would be possible which still would give full bonus and which should behave well or even better in reality.
Would it be possible to change the score-function like this or am I missing something?

bkoep's picture
User offline. Last seen 8 hours 10 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

Yes, you are absolutely right! This is a problem with the score function of the HBond Network Objective that needs to be fixed.

BUNS (buried unsatisfied polar atoms) are only problematic when they are buried. If a H-bond network extends to the protein surface, then the exposed atoms can be satisfied by making H-bonds with water. These exposed polar atoms should not jeopardize protein folding, and they should not incur a penalty.

Joined: 03/28/2020
Groups: Go Science

Thanks bkoep for outlining this point so well. This is exactly what I wanted to express. And many thanks for taking care that this gets implemented. I am really very much looking forward to this. But I guess it will be quite tricky to implement. I wish good luck to the whole team!

Joined: 12/06/2008
Groups: Contenders

I uploaded to the scientists solution "82314 - 2102..." that shows a nitrogen from a network-bound histidine clearly sticking out into solution.

Penalized over 250 points for that. :-(

Looking forward to the Objective fix.

Joined: 12/06/2008
Groups: Contenders

I've got another false BUNS on puzzle 2130 that is costing me over 250 points. See attached photo.

(Looks like we can't attach photos. Go here to view: https://ibb.co/qkHT7GX .)

bkoep's picture
User offline. Last seen 8 hours 10 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

Thanks for the feedback! Can you share this solution with scientists?

Joined: 12/06/2008
Groups: Contenders

Solution "82314-2130" uploaded to scientists. The sidechain in question is GLU 18. One of the oxygens is clearly exposed to solution, yet is counted as BUNS.

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, Boehringer Ingelheim, RosettaCommons