5 replies [Last post]
Joined: 12/06/2008
Groups: Contenders

We now have "new sliders for sidechain and backbone hydrogen bond importance".

What do these do, other than reduce my score? Can someone please explain how they work, and when/why we would use them? The Wiki has little to say about them.

joshmiller's picture
User offline. Last seen 34 min 41 sec ago. Offline
Joined: 09/08/2017
Groups: Foldit Staff
Good question

Hi Boots,

Great questions, sorry for the lack of clarity when releasing this feature.

The new behavior options are simply there to be additional tools for you. Similar to clashing importance, you might find that lowering or raising a subscore's importance may help you get into a better fold. I think some of our scientists said that the hbond subscore importances might be helpful for forming hbond networks or making bonds with a binder target. Ultimately, this is just more power for you to shape the protein as you'd like!

In terms of what tools to use them with, they work exactly like Clashing Importance, so they should affect whichever tools CI affects.

Technically speaking, what all of these sliders do is change the score function used by tools, making tools believe they are making an improvement by weighting the importance of certain subscores differently. However, the Score displayed at the top always uses the unmodified score function, so you will never get an increased score directly by modifying slider values.

So far in my experience, the best way to use them is to crank them up when trying to form hbonds, then you can bring them back down to 1 when your hbonds have formed. The backbone hbond slider might also be useful as a slider for how much you care about your secondary structures keeping all of their backbone-backbone bonds. And the sidechain slider is probably useful when trying to make more bonds to a target for binder puzzles.

Hope this helps! Good luck with experimenting with them and let us know what you find!

Joined: 12/06/2008
Groups: Contenders
Which subscores are modified?

There are many subscores used when calculating the total score.

Which subscores are modified by each slider?

I assume that increasing the slider value above one more heavily weights that/those subscore(s) relevant to the unmodified subscore values, and decreasing the slide value below one less heavily weights that/those subscore(s) relevant to the unmodified subscore values.

Joined: 09/29/2016
Groups: Gargleblasters
Currently with the HBond sliders...

It's only the "Bonding" subscore. We, as players, cannot see how the subscores are broken up into, as we're given the "dumbed-down" version.

Unfortunately this means that BOTH modify the Bonding subscore, but that both impact it differently. Sidechain slider impacting only Sidechain-to-Sidechain and Sidechain-to-Backbone bonds, and Backbone slider only impacting Backbone-to-Backbone bonds.
But yes, the weights work as you said. The end result, though, works in the opposite way that Clashing does. Where Clashing will compact your protein the lower your make it, these HBond sliders will have the tendency to compact it the higher you make them. As bonds that might be weak, will in turn get stronger by bringing stuff closer together.

Sorry about the Wiki not having this info yet, I'm still working on the page. (It'll be at this page when it's done: https://foldit.fandom.com/wiki/Behavior_Menu)

LociOiling's picture
User offline. Last seen 10 hours 48 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
two trains running

I see the Behavior Menu page is now up.

We also have Behavior Options.

Maybe we need to consolidate, we'll see.

Meanwhile, there's also Sidechain H-Bonds and Backbone H-Bonds describing the sliders in detail. Those pages could probably be consolidated too, but it's just easier to take small steps.

Joined: 03/28/2020
Groups: Go Science
I agree that more

I agree that more experimenting with these sliders/LUA functions is required. With your explanation and the wiki I understand now what they are meant to be used for and I am motivated to try them out. Thank you all! But in addition to this, I would appreciate a function in LUA to detect if a HBond has actually formed. Or in general, if a sidechain has HBond or not or how many. Maybe this can be used to automate the creation of HBonds and improve solution quality in conjunction with the slider-functions. This feature was requested by jeff101 some time ago and I think this is a very good idea! I do not see that there is a function to reliably (or at all) do this at the moment.


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