Mutate broken on monomer

Case number:845799-2004896
Opened by:Skippysk8s
Opened on:Friday, March 16, 2018 - 19:27
Last modified:Friday, April 27, 2018 - 20:02

The primitive monomer puzzle keeps inserting hydrophobes sticking out into the water side of the sheet plane. I've tried the mutate button, Rav3n mutate by layers, hand mutating, and several other programs. It keeps mutating all the sheet segments to hydrophobes and the score goes up. This is happening on the sheet plane but not the "hot dogs". Something is off with part of the scoring system. No beta bulges, but clearly not the right protein

(Fri, 03/16/2018 - 19:27  |  7 comments)

Susume's picture
User offline. Last seen 3 hours 36 min ago. Offline
Joined: 10/02/2011

This was true in the original primitive monomer puzzle as well:

So it is not a result of the new scoring algorithm, but of something peculiar in the primitive puzzles.

bkoep's picture
User offline. Last seen 2 days 17 hours ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

Thanks for the note Skippy! We do not necessarily expect these "Primitive" design puzzles to behave well. We've made a lot of changes to Foldit over the years to improve how design puzzles behave; now that we have a new score function, we want to see if all of those changes are still necessary.

It's important that we can see exactly how these puzzles are misbehaving. So anything you can do to find a high-scoring solution (even if you can tell it's a poor protein design) helps us to improve the score function.

For example, previously we had to tweak the old score function so that surface hydrophobics score poorly. Your feedback suggests that the new score function has the same problem, so we will probably need to tweak it again in the same way.

jeff101's picture
User offline. Last seen 3 hours 50 min ago. Offline
Joined: 04/20/2012
Groups: Go Science

I've noticed odd behaviors myself on recent Design Puzzles.
When I put several beta-strands side-by-side to make a sheet
and tried to have one surface of the sheet be hydrophobic and
the other surface be hydrophilic, mutate gradually replaced
many hydrophilic residues on the exterior surface with
hydrophobic ones. This seemed odd to me. Nevertheless,
says the following:

Large aromatic residues (tyrosine, phenylalanine, tryptophan)
and β-branched amino acids (threonine, valine, isoleucine)
are favored to be found in β-strands in the middle of β-sheets.
Interestingly, different types of residues (such as proline)
are likely to be found in the edge strands in β-sheets,
presumably to avoid the "edge-to-edge" association between
proteins that might lead to aggregation and amyloid formation.

According to
Foldit colors tyr, phe, trp, val, & ile orange as for
hydrophobics while it colors thr blue as for hydrophilics,
shows that tyr & thr have -OH groups on their sidechains
while trp has an -NH group on its sidechain, and these
groups could potentially form hydrogen bonds with water.
Perhaps designing sheets with tyr, thr, & trp on the
exterior surface and phe, val, & ile on the interior
surface will help things behave better. If one colors
using Hydro, there will still be many orange residues
on the exterior, but if one colors using CPK instead,
perhaps there will also be more red (O) & blue (N)
residues on the exterior.

LociOiling's picture
User offline. Last seen 1 hour 57 min ago. Offline
Joined: 12/27/2012

WBarme1234 reported that mutate is not working on Beginner Puzzle Symmetry Design.

I'm able to get the puzzle to mutate, but the results are not as expected.

Here's one thing I tried:

1. from the start, make a large helix
2. align the helix parallel to the two shadow copies
3. selection interface, select all, mutate to asparagine ("n")
4. use mutate all ("y") for 3-4 cycles

I'd expect to see hydrophobics on the surface at the end, but the result is that the asparagine is mostly replaced with isoleucine. The sequence I end up with is


The isoleucines all show as exposed. There's only a single hydrophilic, a threonine at segment 34, and it's toward the inside of the three helixes.

All this is on 20180416-5eaeddec33-win_x86-devprev. I'll apply the latest devprev and try again.

LociOiling's picture
User offline. Last seen 1 hour 57 min ago. Offline
Joined: 12/27/2012

Repeating the experiment on 20180426-5042fb62a9-win_x86-devprev ended with much the same results. The threonine moved to segment 17, but it's still pointing in.

LociOiling's picture
User offline. Last seen 1 hour 57 min ago. Offline
Joined: 12/27/2012

I tried a different experiment using a good midgame solution from 1513.

The steps were:

1. Selected all, mutated to asparagine
2. Mutate all

The result looked reasonable, no exposeds, and score that was at least close to the starting point.

Looks like more research is needed....

bkoep's picture
User offline. Last seen 2 days 17 hours ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

Thanks LociOiling!

In the past couple of weeks, we've reintroduced some tweaks to the score function to help with exposed hydrophobics. So it's good to hear that Puzzle 1513 is behaving "as expected." We have not yet updated the Beginner puzzles with these tweaks, so it makes sense that those puzzles still suffer from this problem.


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