Back to foldit 101
Foldit offers two levels of puzzles that are played for points. New players who have completed intro puzzles can play beginner puzzles, and all players can play regular, non-beginner puzzles, which are marked as "intermediate" or "advanced". (See gameplay for more on the ins-and-outs of playing Foldit.)
There are several types of non-beginner Foldit puzzles. Each type of puzzle has different scientific goals and calls for different strategies.
As of October 2017, the most common types of Foldit puzzles are:
- design - players design a new protein, including the amino acid sequence, secondary structure, and overall shape
- de-novo - given a fixed sequence and a suggested secondary structure, players try to determine the overall shape
- revisiting - players try to solve previous puzzles
- symmetry - players try to design a protein that binds with one or more copies of itself
- sketchbook - players try to fold a protein using a limited number of moves
- small molecule design - players design small non-protein molecules that bind to an existing protein
- multi-start - multiple starting poses are available, players try to find the one that produces the best overall shape - this long-dormant category has recently been revived
In some cases, puzzles include additional information derived from experiments. Players can use this information as a guide to folding. These types of puzzles include:
- electron density - an electron density "cloud" is shown as a guide to folding
- contact map - the contact map shows which segments are likely to be "in contact" with each other
In addition to regular design puzzles, which usually involve designing a protein from scratch, Foldit offers "docking" design puzzles, which have usually focused on particular disease targets, such as the HIV and Ebola viruses. Docking puzzles usually involve designing a small section of protein to attach to or "dock" with a larger section of a virus or bacteria. Usually most of the disease target is locked, preventing it from changing shape.
For the Foldit Aflatoxin Challenge, the first puzzle, 1440: Aflatoxin Challenge: Round 1, has a different approach. The puzzle has a large protein known to act as an enzyme. An aflatoxin molecule is loosely bonded to the large protein. The goal is to improve the bonding between the protein and the aflatoxin. Much of protein is locked, but sections of the protein near the aflatoxin can be designed. In some ways, this is the opposite of the small moelcule design puzzle.
There are also several types of Foldit puzzle seen in the past:
- GUI hand-folding - players are limited to handle folding and simple GUI recipes
- quest to the native (QTTN) - a translucent guide is shown, players can use hand folding and bands to match the protein to the guide
- centroid - sidechains aren't shown, each segment is seen as a fuzzy ball, players try to find the best overall shape
- alignment - multiple sequences are available in the alignment tool, players may align all or part of the protein to a given sequence to shape the protein
- exploration - an "exploration multiplier" affects the score, see "The story behind solution diversification (or exploration) puzzles"
- CASP - CASP is a scientific competition between protein-folding labs, held every two years. Foldit last participated in CASP 11, held in 2014, and sat out CASP 12 in 2016.
These puzzle types may or may not reappear in the future.
Description of puzzle types
While the goal is always to produce a well-folded protein, each puzzle type requires a different approach, and may have a different scientific goal.
Typical recent design puzzles allow players a specific number of segments (or "residues") to design. In recent puzzles of this type, players may delete unwanted segments without a penalty, but receive a penalty for adding segments beyond the original number.
This type of design puzzle starts with all segments in a straight "extended chain". All segments are the same amino acid type at the start, and no secondary structure is defined.
Design puzzles of this type typically have multiple conditions, which affect scoring. For example, there's typically a "core existence" condition, which requires a certain percentage of residues to be "buried in the core of your design". (The "core" is the tightly-packed center of the protein, and "buried" means that the segment's sidechain is where water molecules can't get at it.) With the core existence condition, you receive a penalty if the goal isn't met, and bonus if it is.
The main scientific goal of the common type of design puzzle is to create proteins that fold up on their own the way that natural proteins do. The concept of "ideality" has been important in improving designs. Previous generations of proteins designed by Foldit players have been "non-ideal" in various ways. Successive changes to the game have attempted to bring Foldit results more in line with "ideal" results.
Recent Foldit design puzzles have included an "ideal loops" condition, which penalizes designs where the "loop" sections between sheet and helixes depart from ideal shapes. See ideal loops for more details.
hydrogen bond network design puzzles
One type of design puzzle places special emphasis on "hydrogen bond networks" that form between hydrophobic residues on the surface of protein. In these puzzles, a condition rewards hydrogen bond networks that meet certain criteria. In symmetry puzzles, hydrogen bond networks between the symmetric units are considered especially important.
target design puzzles
In addition to the "extended chain" design puzzle, Foldit has offered smaller "target design" puzzles. In these puzzles, the goal is typically design a small protein fragment that binds to a larger protein. The larger protein usually has a more or less fixed shape, although some sections may be able to move. Some or all of the larger protein's sidechains may be moved. Although the part to be designed is small, target design puzzles are probably the complicated type of Foldit puzzle.
symmetry design puzzles
Most symmetry puzzles are an extension of a design puzzle. See the description of symmetry puzzles for more.
Like design puzzles, de-novo puzzles typically start with a straight "extended chain". Unlike design puzzles, the amino acid type of each segment in the chain can't be changed. (In Foldit terms, the segments aren't mutable.) De-novo puzzles usually start with a secondary structure predicted by PSPIRED, see secondary structure prediction tools for more on how this works.
A de-novo puzzle normally represents protein that has no known, or at least no published, solution. Some de-novo puzzles represent proteins which are about to be published. The scientific goal is to see how well player solutions match the published results. As of January 2017, many recent de-novo puzzles represent Foldit player designs from previous design puzzles. The goal for these puzzle is to evaluate how players predict the structure of these designed proteins as compared to automated systems.
Revisiting puzzles present previous puzzles. The proteins in these puzzles generally have published solutions. The scientific goal is to evaluate how changes to Foldit have affected the game. One notable change in recent revisiting puzzles has been the introduction of a disulfide bridge condition that awards bonding between two cysteines. Disulfide bridges are an important factor in protein stability, but historically, Foldit scoring hasn't reflected their contribution.
Most symmetry puzzles are an extension of a normal design puzzle. Players design a single chain of segments, as in a normal design puzzle. The single chain is called a "monomer". In a symmetry puzzle, there are one or more shadow copies of the monomer, that change as the player changes the monomer. The goal is to produce not only a good monomer, but one which binds to one or more copies of itself. Recent symmetry puzzles include both a "core existence: monomer" condition, similar to the normal core existence condition, and a "core existence: complex" condition, which rewards bonding between the copies of the monomer.
A regular design puzzle will have "monomer" in its title, as in "85 Residue Monomer Design". A symmetry puzzle will have "symmetric" in the title along with the type of symmetry, for example "70 Residue Symmetric Hexamer Design". Just as "monomer" means one copy of the chain, "dimer" is two copies, "trimer" is three copies, "tetramer" is four copies, "pentamer" is five copies, and "hexamer" is six copies. If you're still curious, a "heptamer" would be seven copies, but so far, no Foldit puzzle has tried a heptamer. Just for reference, the "complex" of monomers is also called an "oligomer". Just for the sake of confusion, what we're calling a "monomer" is technically a "polymer"....
electron density puzzles
Electron density (ED) puzzles present a "cloud" of electron density information derived from experimental results. The cloud can be visualized in various ways. With the correct settings, players may be able to see the backbone and sidechain positions of the protein in the cloud. The sidechains of the "ring" amino acids, including tryptophan, tyrosine, histidine, and phenylalanine, are usually the easist feature to spot.
Players can place notes on the ED cloud, and may also make bands from the protein to the cloud.
In an electron density puzzle, each segment has a sub-score called "density" that measures how well the segment is covered by the density cloud. A higher score is better, but it doesn't necessarily mean that the segment is the the correct part of the density cloud. Recipes can access a segment's density sub-score and try to improve it.
contact map puzzles
Like electron density puzzles, contact map puzzles have additional experimental information available. In a contact map puzzle, this includes a table which indicates whether two segments are "in contact". Being "in contact" just means the two segments are reasonably close to each other, not that they are necessarily bonded in any way. The contact between two segments may be represented as either a probability (a real number between 0 and 1) or a yes/no choice (0 for no contact, 1 for contact). A contact bonus condition applies a bonus for each match between the protein and the contact map. The contact map bonus may be weighted according to the contact's probability, with more probable contacts receiving a higher score.
The contact map information is presented in two ways. There's a "Contact Map" window available in the Foldit user interface. The Contact Map is a graph with the segment number as both the x and y axis. Since a segment is always in contact with itself, the diagonal [1, 1], [2, 2], [3, 3] and so on is a solid line. The two sides of this diagonal are mirror images of each other. A dot on the contact map means the two segments may be in contact, based on experimental results. The coloring of the dot indicates the probability of the contact.
In addition to the visual presentation, recipes can also retrieve the contact map information.
small molecule design
The small molecule design category was formerly referred to as drug design. This type of puzzle has been in the works for a while, but new puzzles of this type finally started to appear in September, 2017.
In small molecule design, player can modify a small molecule to make better bonds with a existing protein. The protein may be mostly locked.
These blog posts give more background on this type of puzzle
- Foldit drug design introduction, January 2015
- Foldit Drug Design Part Two, February 2015
- Foldit Drug Design Blog: Interface Update, March 2015
- Let's Get Ready for Drug Design Puzzles!, September 2015
- Big Update: Experimental Client (Drug Design), May 2016
Notes on gameplay
As always, Foldit is a combination of science and game. There are many different ways to play the game of Foldit. Here are some notes on the game and how to play it. This information is included here because each of these "ways to play" can also be considered a "type of puzzle".
Most puzzles run for about a week, although beginner puzzles run longer. Some non-beginner puzzles are considered especially difficult, and may run for two weeks or longer.
Players are ranked by score when each puzzle ends, and receive "global points", which contribute to their Foldit ranking. Typically, beginner puzzles are limited players with less than 150 global points. (Players with more global points can play beginner puzzles, but they won't receive points.)
At the start, players play on their own, as "soloists". Players can join a group, are then eligible to play as "evolvers". An "evolver" works on a shared solution that another soloist or evolver in the same group previously worked on. At the end of the puzzle, evolvers are ranked and scored just like soloists. Additionally, groups are ranked and scored. Each group's best solution from either the soloist or evolver side is used to determine group ranking.
Foldit also offers various "contests", which typically run for much longer times than puzzles. No points are assigned for contests. See the contests page for the current contests. Unlike regular puzzles, players must register for a contest.
Some puzzles don't result in points. In some cases, puzzles are closed early due to technical problems. Players receive no points for these puzzles. Typically, a puzzle closed for technical reasons will be reopened under a new name a short time later, and players will be able to load their work from the problem version of the puzzle to the new version.
There can be multiple versions of the Foldit client, each in its own "update group". The default update group is "main", and puzzles in this group normally receive points. There's also a "developer preview" update group called "devprev", and an "experimental" update group for work related to drug design tools. In devprev, puzzles with "[DEVPREV]" in the title are typically long-running, and receive zero points. In experimental, puzzles with with "[e]" in their titles also receive zero points. These devprev and experimental puzzles are intended mainly for testing purposes, and players are encouraged to try them.
Players sometimes share screenshots of completed puzzles in chat. You can see galleries of these images for Puzzles 1001 and on and Puzzles 1 to 1000. Especially for design puzzles, images of previous puzzles may offer important clues as to what works.
This page underwent a major update in January 2017. This section contains some of the previous content for reference. The content in this section will likely be moved elsewhere at some point.
Foldit players get the opportunity to engage in some of science's current puzzles; the mechanics of the flu, carbon dioxide fixation and of course the Critical Assessment of protein Structure Prediction (CASP) experiments, which aim at establishing the current state of the art in protein structure prediction.
To quote Foldit's description:
- CASP (Critical Assessment of Techniques for Protein Structure Prediction) is a protein structure prediction competition that is held every 2 years, with labs from around the world testing out their newest state-of-the-art folding methods on unpublished structures.
The nature of CASP target proteins has changed over the years. Foldit last participated in CASP 11, held in 2014. Foldit did not participate in CASP 12, held in 2016. Foldit Participation in CASP 13, scheduled for 2018 is yet to be determined.
The results for Foldit players and groups are recorded in the Foldit German wiki CASP results page
Secondary Structure (SS)
Sheets, helixes (helices) and loops – most puzzles show the predicted secondary structure, and although you can change these using Structure Mode, usually folders are asked to stay close to the predicted secondary structure.
Also - at your own risk - you may find Foldit-aas reference useful
Some puzzles are made up of one large and one or more smaller parts, try to make them bond well together. Puzzles may only provide a section or fragment of a protein to focus on the area of interest of the puzzle design. Usually design or mini puzzles but you may also find this in a standard puzzle.
Tip: Google Ligand or Paratope for explanations of these small pieces and how they may work.
Warning: Some recipes may not work well on multipart puzzles.
Thin red lines may appear between parts of the protein, these are Constraints and severely restrict the shape of the puzzle to meet the needs of the puzzle design, constraints can be stretched but not broken. Usually the puzzle will feel stiffer and slower. Constraints are not often used on beginner puzzles.
Useful tips for dealing with Constraints from an advanced player.
In design puzzles which contain a ligand (shown in brown)there are constraints (shown in green) to limit the movement of the ligand. Check the show constraints box in the View menu to make these visible.
There are three levels of puzzle. Once you have completed the tutorials you can explore and compete on any puzzle. (Some challenges and contests may be restricted to a specific group of folders.) Puzzles are usually 100 points each, some smaller puzzles may be 50 points and the grand designs can be worth 150.
Beginner Puzzles <150
Some puzzle have "<150" in their name. The puzzles are exactly like science puzzles and the whole toolset is available, the restriction is based only on the number of global points you have. For these puzzles, you can click in the header to find help in the wiki. Foldit 101 suggests that you take advantage of building your points by playing the <150 while you can.
These puzzles are the easier science puzzles. The protein can be short or the modification to make less important than in other science puzzles.
Intermediate and Advanced Puzzles
These are progressively more difficult and usually require a good understanding of the tools. The term ‘easy’ should be treated with caution; relative ease is based on a comparison with CASP targets.
Quest to the Native (QTTN) or Guide Puzzles
Are usually beginner puzzles, these show a grey translucent guide (which is the native shape for the protein), and you need to fold the protein to match the guide. Folders will usually use freezing and bands to align the backbone and sidechains into shape.
Align the whole guide by clicking the Align Guide button in the Actions menu or pressing G on the keyboard on old interface, A on selection interface.
Tip from Beta Helix
By right clicking on a region of the protein (not the guide) it will pull up a menu with 6 tools. The bottom right one is "Align Guide". Clicking on it will align the native guide to the region of the protein you selected.
This will help you see what sections on your protein correspond to the guide!
Working with the guide is much more user friendly when you use the advanced viewing options.
In the game click on Menu->General Options and select "Show Advanced GUI" and click Done.
Click on View, turn on 'Show Outline,' set the background color to white, and view stubs on the sidechains, view the guide, and set color relative to guide. Turn all the other view options off."
Tip from CFC
You may find it gets considerably easier to band individual chains to a guide if you do a couple of things. Shift A will bring up all chains as visible, then set 2 or 3 bands just into space on every chain in the section you want to work on, then Shift D. Now, only the chains with bands will stay visible. Put the cursor on the first chain you want to work on, and press Q, and the whole protein will shift so that this chain is now at the centre of the universe, when you rotate the background by dragging. (If you put the cursor on the background and press Q, it'll recentre the whole protein, like Home key - the solution we see in #global when people lose their visuals off-screen). By now rotating the view, and dragging each band tip, you can now match the guide, chain by chain slightly more easily.
Tip: Read Tlaloc's Strategy for Guide Puzzles.
Tip: Grom's video
Tip: drumpeter video
Are a standard puzzle without a guide, some may have constraints. May be any puzzle level.
De Novo Freestyle Puzzles
You start with a single extended chain, often there is a full or partial predicted secondary structure, simply fold it to the highest score you can, using rebuilds to curl the helixes. Generally these are advanced puzzles. Some tips to get started: How_to_start_folding_a_DeNovo_(freestyle)_puzzle?, Hand folding de novo S pletsch, marie_s strategy for hand folding and some videos.
Docking puzzles : They are usually partially or completely locked (segments showing in a solid grey) with a small area which can be manipulated, sometimes just the sidechains move and or mutate. Design Mode is available in the toolbar; some puzzles allow insertion of segments and all design puzzles will allow mutation of certain segments. It is not uncommon in these puzzles to see ten or twenty players with the same score integer, separated by decimals only.
Best designs will be selected by the researchers, tested on Foldit@home and then synthesized in the lab.
Symmetry puzzles : Often a de-novo sequence, you cannot interact directly with the other components and they will be exactly the same as the active chain.
Design puzzles typically have one or more conditions.
Electron density puzzles were introduced in June 2011. Electron Density Puzzles page
Foldit science blog post and comments here: http://fold.it/portal/node/989898 all levels.
Were introduced in 2011. There are two scores running in exploration puzzles, the stability score, (like a standard puzzle score) is multiplied by the exploration multiplier; which varies up to 3 based on how much you have explored away from the original structure; the result gives you the credit score. See also "The story behind solution diversification (or exploration) puzzles".
Tip: watch Black Belt Folding Electron Density video .
Warning: Because the scoring is different most recipes do not score properly on these puzzles, several new recipes have been written for Exploration puzzles.
Multi-Start or Multi-Template Puzzles
Use the Alignment Tool to view each of the provided templates. Depending on the puzzle, the templates may be very different and score accordingly. Usually intermediate or advanced level puzzles.
Tip: Use the Tracks function to put each template in its own track when you thread it, you can then easily revisit and work on different templates later.
Tip: see beta_helix's alignment tool notes
Focus on small fragments of proteins, usually less than 40 amino acids (aas) and are often multipart. These are very fast to run and easy to experiment with. All levels.
In Centroid puzzles the sidechains are not visible.
A range of contest puzzles are available, you can also create your own contest based on a range of preset options. Contest puzzles help you hone your skills when other new puzzles are not available. Have fun!
Challenge puzzles are like contest puzzles, but are run as a contest between 2 or more people to see who can get the best score.