puzzle picture
1432: HIV Protease Inhibitor Small Molecule Design
Status: Closed

Summary

Name: 1432: HIV Protease Inhibitor Small Molecule Design
Status: Closed
Created: 09/19/2017
Points: 100
Expired: 09/27/2017 - 19:00
Difficulty: Intermediate
Description: This is the first official small molecule design puzzle. The goal is to change the ligand in the center of the protein to find better inhibitors of HIV protease. Use the Ligand Design tool in the selection interface (or the "Ligand Design" mode in the original interface) to alter the chemical identity of the inhibitor. See the blog post for more information on the puzzle and small molecule design.
Categories: Experimental, Small Molecule Design

Top Groups

RankGroupScorePoints
1Go Science20,392100
2Beta Folders20,29980
3Contenders20,23763
4Gargleblasters19,72949
5Hun-Magyar Csapat17,99037

Top Evolvers

Top Soloists



Need this puzzle? Log in to download.  

Comments

frood66's picture
User offline. Last seen 10 hours 39 min ago. Offline
Joined: 09/20/2011
Groups: Marvin's bunch
puzzle name

please call this what it is? This is a drug design puzzle - or so it would appear.

rmoretti's picture
User offline. Last seen 10 hours 34 min ago. Offline
Joined: 01/15/2010
Groups: None
"Small Molecule Design"

We're calling this category of puzzle (one where you modify the chemical identity of a non-protein ligand) "Small Molecule Design".

"Drug design" also works. But "small molecule design" is a little more general, as the molecules we're designing aren't necessarily drugs per se. (Also, protein-based and RNA-based drugs do exist.)

frood66's picture
User offline. Last seen 10 hours 39 min ago. Offline
Joined: 09/20/2011
Groups: Marvin's bunch
well - I only use straight

well - I only use straight language.....this is a drug design....look at it as u will

LociOiling's picture
User offline. Last seen 5 hours 43 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
design queue - no progress indicator

It looks like the design queue tool has an interface similar to a manual remix, but the bar underneath the navigation arrows doesn't show "design x of y". (Or anything, actually). This makes it impossible to know how many designs are in the queue.

LociOiling's picture
User offline. Last seen 5 hours 43 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
Small Molecule Design category page gives "access denied"

The category page for Small Molecule Design gives "access denied".

rmoretti's picture
User offline. Last seen 10 hours 34 min ago. Offline
Joined: 01/15/2010
Groups: None
Try now

Sorry about that - it was set to private.

LociOiling's picture
User offline. Last seen 5 hours 43 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
Small Molecule Design category page gives "access denied"

The category page for Small Molecule Design gives "access denied".

SaraL's picture
User offline. Last seen 1 day 11 hours ago. Offline
Joined: 09/20/2011
Groups: Gargleblasters
did something wrong?

Not familiar with the editing process, so I went and changed chemical bonds and swapped out individual atoms, while using the "similarity to known binders" score as a guide. I got up to 40% and couldn't get it any higher.

I attempted to upload my solution to share with my group, but I got a messages saying "Error in sequence."

Back to the drawing board?

Joined: 09/24/2012
Groups: Go Science
same error in sequence

I got the same error for some solutions (unable to load or share a saved solution), also with high bonus (>50% similarity) and same method.

See Feedback here:
http://fold.it/portal/node/2004221

Moreover, the related client crashes after some time.

For another solution (with lower bonus), there was no problem.

rmoretti's picture
User offline. Last seen 10 hours 34 min ago. Offline
Joined: 01/15/2010
Groups: None
Similarity isn't everything

There's going to be a limit to how high the "similarity to known binders" will get. You can use that number to help you guide your designs, but we're also interested in what novel designs you can come up with. If your similarity score tops out, switch your focus to something else (like the main score) and try to increase that.

Regarding the shared solution issue, we'll take a look into it: see also this feedback entry - If you have any additional info about the sharing issue, feel free to add it there.

Joined: 05/19/2009
Groups: Contenders
Where are the introduction puzzles ?

1. Where are the intro puzzles for small molecule design ?
2. Is it correct that none of the graphics have any function ?
3. What is it that we are supposed to do with this ?
4. What scores well, what doesn't ?
5. What are the controls for any functions or is everthing click and drop based ?
6. How do we play these puzzles ?

rmoretti's picture
User offline. Last seen 10 hours 34 min ago. Offline
Joined: 01/15/2010
Groups: None
Early days

Keep in mind that it's still early days with the small molecule design puzzles. We're still working out how best to enable the effective design of small molecules in Foldit.

1) We're still working on the intro puzzles for the small molecule design features. If you have suggestions on how to approach making an effective intro puzzle for these tools, please let us know.

2) If you're referring to the Ligand View tools, none of those are going to change the structure (and hence your score) - they're purely for getting more information out of the structure. You hopefully will be able to use that information to inform your design strategy, and to inform which modifications you make.

5) If you're referring to recipe scripting, the Ligand Design functionalities aren't necessarily exposed yet. One complication is that the Ligand Design tools work on an atom-by-atom basis, versus other tools in Foldit, which work on a residue/segment-level basis. How to effectively work selecting particular atoms into the recipe scripting system is an open question. -- If there are any recipe writers out there who have ideas, please let us know.

3 & 6) The main activity which you'll be performing during a small molecule design puzzle is using the Ligand Design tools to change the chemical identity (how the bonds and atoms are connected) of the central ligand to make new and better scoring ligands. You'll probably also want to use the other Foldit tools like shake and wiggle to optimize the ligand into the structure and the structure around the ligand. A successful small molecule designer will likely use all the tools at their disposal to optimize the structure.

4 & 6) As a new puzzle type, the best strategies for approaching these puzzles are something you (as part of the Foldit player community) have to figure out. I can't necessarily say what approaches are going to work out - that's as much a mystery to me as it is to you. Like the early days of protein folding, where Foldit strategies needed to be developed, the precise approach one should take on small molecule design puzzles is something to be worked out.

That said, all the principles from protein folding should also apply here. Hydrophobic (carbon/hydrogen rich) sections of the ligand should be buried with hydrophobic sidechains. Polar (oxygen/nitrogen righ) sections of the ligand should either be exposed or (preferably) be making hydrogen bonds to the sidechains and backbone of the protein. Polar sidechains in the ligand binding side should be paired up with polar groups on the ligand, such that they make hydrogen bonds. Good packing of the ligand/protein interface is also important, so fill voids without making clashes. (Also keep in mind that the protein also contributes to the score, so don't make the protein unhappy to better fit the ligand.) The wrinkle for ligand design versus standard protein design is that you have relatively free rein to change the identity of the ligand to accomplish this.

As I said, it's still early days, and it's going to be a learning experience for both sides. I completely anticipate that we (as developers) will learn things from these early small molecule design puzzles which will change how we set up these puzzles, to make it easier for players to achieve success. (This is similar to the protein design puzzles, where there's been multiple iterations of puzzle settings, and new tools introduced to help with that process.)

LociOiling's picture
User offline. Last seen 5 hours 43 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
selection mode may help...

As recommended in one of the blog posts, it's easier to use selection mode. Also, the "Cartoon Ligand" view option and "Ligand Specific" coloring option may help.

In selection mode, you can select the ligand, segment 199. (Ligand-specific coloring makes it stand out.) The ligand tools then appear along the bottom of the window.

The "Design Ligand" option, or shortcut L, initially has all its sub-options greyed out. Click on one of the atoms of the ligand to select it. The selected atom is highlighted with a halo, and the valid options are no longer greyed out.

The exact options depend on the atom. If your atom is eligible for one of the larger molecules (CO2 through benzene), hovering over the option gives you a preview of what things would look like. (Hint: carbon has more options.) No previews for simple atoms, however.

The other options seem to do something as well, and so far no crashes. The Ligand Queue tool is a bit confusing, and it's not clear what MMFF Wiggle does, but it's good for losing points.

It can be a little hard to tell which atom you've selected. The CPK colors are changed due to having the whole ligand selected. Otherwise, nitrogen is blue, oxygen is red, sulfur is yellow, hydrogen is white, and anything else is carbon. The Small Molecule Color under Ligand Properties lets you change the carbons to yellow if you want, but it's a slightly brighter yellow than sulfur's yellow.

Atom selections are cumulative, so when you click a second atom, the first stays selected. You can click the first atom again to deselect it.

It's easy to deselect the ligand if you click in the wrong spot.

Atom choices aren't always valid, even when they're not greyed out. If you make a bad pick, the error message tends to hide behind the options. (Moving the ligand up the screen may help.)

Aside from that, it's just a simple matter of biochemistry....

Joined: 09/22/2017
Groups: None
I'd love to see an import

I'd love to see an import SMILES function added

Joined: 05/19/2009
Groups: Contenders
Thank you

Thank you for the explanations, I am making progress now.

Joined: 05/19/2009
Groups: Contenders
Error, please see feedback.
Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Shake is weird in undo graph

Shake seems to produce a stop on the undo graph for every sidechain that moves, rather than one stop for the entire shake. I shook for over 3 minutes, and the iteration counter still said 1 but there were 28 blue stops produced on the undo graph.

Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Double negative

"Ligand is very unlikely unsynthesizable." Total UNs in this sentence: 2 is too high!

Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Clicking ligand atoms should affect the one in front

When trying to select or deselect a ligand atom, if there is another atom immediately behind the one I am clicking, the game consistently highlights (or de-highlights) the one behind instead of the one in front. When I turn the view so there is no atom behind my target and then click in the same way, it highlights/de-highlights the one I want.

Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Numbering scheme for heavy atoms

Is there a numbering scheme for the heavy atoms that would remain consistent as we add and remove atoms? It's very hard to keep notes or remember what you tried when you can't identify an atom as the same one you changed last time.

Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Why do the Nitrogens have 4 bonds?

Basic chemistry time: I think I understand why each carbon has 4 atoms hanging off it (less if some are double bonds) and each oxygen has 2. But why does each nitrogen also have 4? Seems to me nitrogen should always have 3 (or 2 if there's a double bond).

http://fold.it/portal/files/chatimg/irc_341210_1506461694.png
Two nitrogens with 4 bonds each, an oxygen with 2, and the carbons have 4.

rmoretti's picture
User offline. Last seen 10 hours 34 min ago. Offline
Joined: 01/15/2010
Groups: None
pH-dependent protonation

We try to model the compounds at physiological pH. (That is, at the pH one would encounter in the body.) At this pH (around pH 7 to pH 8), most amine-like nitrogens are going to be positively charged with an extra hydrogen hanging off of it, for a total of four bonds.

If you enable showing hydrogens on the protein - even on non-small molecule design puzzles - you'll notice that the sidechain amine nitrogen of lysine is also bonded to four atoms for the same reason. The backbone nitrogen is indeed only bonded to three other atoms, but it's an "amide" nitrogen, not an "amine" nitrogen, due to being right next to the carbon-double-bond-oxygen carbonyl group of the next residue segment. The ligand design code in Foldit tries to detect what sort of nitrogen is on the compound, and then put the correct number of additional hydrogens onto it, based on what would happen in a physiological pH.

Susume's picture
User offline. Last seen 6 hours 53 min ago. Offline
Joined: 10/02/2011
Thanks!

Thanks for the quick reply. I will have to google around and see if I can find some beginner-level reading that covers this.

I notice that the small molecule design lets me hang 4 heavy atoms off a nitrogen, or 3 heavy atoms with a double bond to one, or 2 carbons with double bonds to each. When I attach an oxygen to the carbon next to the nitrogen, it sometimes reduces the nitrogen's bonds to 3, but not always. At this point I am poking around randomly to see what happens; I guess foldit will get it right.

Meanwhile when I try to google "amine-like nitrogen," google thinks I want to read about "anime-like nitrogen." I guess that's a thing.

markm457's picture
User offline. Last seen 6 days 18 hours ago. Offline
Joined: 11/16/2012
Policy for scoring flawed puzzles

Last November, bkoep wrote the following comment about the Foldit policy on scoring of flawed puzzles:

http://fold.it/portal/node/2003010#comment-33734

I've heard various rumors. Can we have clarification that this policy applies to the flawed puzzle 1432 and that it will be awarded zero points?

Joined: 06/08/2008
Groups: Go Science
Lab Results from Ligand Puzzle

I cant wait to see the results from the puzzle.

Lets hope our efforts will be fruitful

Great Team Work Everyone !!!

great start to the Battle of the Designers

Get Started: Download
  Windows    OSX    Linux  
Windows
(Vista/7/8)
OSX
(10.7 or later)
Linux
(64-bit)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Search
Only search fold.it
Recommend Foldit
User login
Soloists
Evolvers
Groups
Topics
Top New Users
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, Microsoft, Adobe, RosettaCommons