My preference was to dismiss the writing of this article, as several very good methods are already present on this wiki, however having been asked, I resort now to disclosing a step by step analysis of the means by which I fold proteins.
Early Game Edit
To me this portion of the game is one of the most significant, as failing to achieve a good start at this point will severely hamper your ability to make important gains later. There are a couple of techniques which I find help with this significantly. At the very beginning of the puzzle, this is the starting technique.
- Wiggle until the score doesn't move for a while
- Repeat steps 1 and 2 until no further gain is made
- Reset the puzzle
- Wiggle until the score doesn't move for a while
- Repeat steps 5 and 6 until no further gain is made
- Restore Best.
The above steps will in many cases mean a difference of 100 points or more to the starting point of a puzzle. With some puzzles I have found the difference to be as high as 400 points. Another good step to add in there would be after seeing where these two end up, is to do a third pass. If for example one yielded a score of 10,352, and the other a score of 10,425, I would, begin again with the better of the two, stopping at about 10,100, shaking, wiggling, stopping again at about 10,200, shaking again and continuing on stopping and shaking about every 100 points or so. Occasionally this offers a better starting position than the other other two methods.
After having found the starting structure with which we will be working, the next step is generally to do an inspection of the puzzle. Make a mental note of areas which stand out as being low yielding. Also look for portions of the puzzle which appear to be out of place. The following is a list of things which I look for:
- Ends of protein, You will find that on many of the puzzles, in the optimal configuration both ends of the protein often appear right next to each other, often parallel for several segments before the end, in others they are still parallel, yet pointing off in opposite directions. If the ends are in random locations, consider that you may not have a very good starting structure, and perhaps take a look at others, or repeat the steps in the first section again. This is not an absolute (as far as I know) so don't base your whole strategy on this, just keep it in mind
- Areas with multiple Phenylalanine and/or Tyrosine sidechains. These are the ones with hexagon shape on them, and tweaking them a bit to get the hexagons parallel to each other, often helps immensely.
- Areas with Proline sidechains, as these almost never appear on a helix or sheet (sometimes at the very end of the helix) they are a good indicator of where the different portions of the protein start and end. Though usually not immediately helpful, it's good to get to know your puzzle a bit. Sometimes, this will make the difference on where I decide to try to bend the backbone.
Once the above areas have been examined and perhaps some basic pulls and adjustments made, it's time to move on to the mid game and begin really tweaking this thing.
Mid Game Edit
Mid Game is where all the major points jumps come in, and the amount of time spent on mid game can make huge differences later on. My progression through the mid game portion is generally as follows:
- Smash and Shake: Drag portions of the puzzle into each other, particularly I focus most of this effort towards any helix, pushing it into the surrounding portions of the puzzle. Knowing how far to push is very subjective, typically I do so until both the side of the helix, and the area it was pushed into are red with little stars indicating clashes. Once it's there, shake. Now I tap the W key a couple of times briefly (no more than a second between taps) shake again, and repeat this a couple times, then let the puzzle wiggle out. If an improvement is made, keep it, if not, restore best. I find myself doing this over and over again for 10 - 30 minutes on any puzzle. Basically, keep at it until it becomes frustrating that you are dropping each time, rather than gaining.
- Pull and Shake: After smashing fails to yield anything else, now it's time to try it's opposite, start pulling pieces apart a little bit, shaking and wiggling them back together. If the new score is higher, great, if not reset best. Also, take note of any sidechains which frequently change position while doing this, often I will lock these so that I can come back and play with them a little later.
- Sidechain_Position: After the easy stuff no longer produces additional points, it's time to start manipulating this manually. If on this puzzle, I have locked sidechains from the previous step, this is the starting point for me, I will unlock them 1 by one, and try the different sidechain strategies already laid out in the Sidechain_Position article. After focusing on the locked chains, I typically go after the Tryptophan (hexagon with pentagon), the previously noted Phenylalanine and Tyrosine and the Arginine (very long squiggly one) trying to find better positions for them. If it is not too boring or idiotic sounding, please also see the section at the bottom of this page relating my thoughts on sidechain "realm".
- Just play with things a bit, this is a good time to try out Rebuild, Local_Rebuild_Strategy,Manual_Rebuild, or Tweak. This is also a good time to band pieces together to try to increase Compactness, and to get rid of Constraints. though I should stress, I keep these to a minimum, and typically only do them on extremely troubled areas, or after at least two passes through with the above techniques do not produce satisfactory improvements. I do not like rebuilding a protein unless an area has either persistent constraints, or red areas which do not become fixed by simple direct manipulation. I do like the tweak tool, but believe that it should be used sparingly. (I typically limit it's rotation to no more than about 5 degrees around the axis of the sheet or helix at a time)
It is important to note, any time your score increases, you should pull the structure slightly, not allowing it to drop more than about 100 points, and wiggle it back together again, do this several times until it stops yielding points. Doing this will add 2-5 points to each improvement you gain in most cases. After spending time on the mid game steps, I will typically either quit for the night, take a break, or move on to another puzzle. After coming back to the puzzle from a break it's good to go through the midgame steps one more time. It is important to get everything you possibly can before moving on to end game strategy.
Sidechain Positions Edit
I would like to add to this page a list of observations and trends I've noticed for sidechains.
Arginine, Glutamic_acid, Lysine These are good sidechains to use to cover things. If you have an exposed hydrophobic that cannot be buried no matter what you try, and there is one of these nearby, try covering it, if only partially. If they are not covering something, I find they like to be stretched out completely /\/\/- so they make an even zigzag, then pull the protein a bit, let it wiggle back together and shake to see if they snap back or pick a different position, often they will stay extended.
Isoleucine One of my favorite target to play with. The odd arm can occur in one of 3 positions, 2 down, 1 up. The up position is usually preferred when the sidechain is buried well, one of the two down positions is usually preferred when the sidechain is near the outside of the protein. In addition the base of it spins, and flipping the arm then trying different rotations often yields very different results.
Leucine This sidechain often takes a little playing with, there are 3 positions for the lower half of it, and they can point like a Y. 2 of the positions are toward 1 direction, and the third is in the opposite. Often large gains can be made by putting this in the odd direction. The upper half rotates and it's rotation can significantly impact the direction that yields the best score in the lower half.
Methionine When this occurs at the end of the protein, I will often spend a great deal of effort rebuilding the last 2-3 segments of the structure where it occurs. Internally in a protein it can take many shapes, and burying it can be often difficult.
Phenylalanine, Tyrosine Bury these good, when there are more than 1 in the same area, try to stack them like pancakes. Rotating the hexagon about it's stem 180 degrees (so it looks like no change has occurred, but it's actually flipped over) often yields points.
Proline Only two real positions for this one, but which it is in can make huge impacts. If one of the ends is exposed, it's usually an indicator that area needs to be rebuilt. This sidechain is also a good indicator of where a helix starts or ends. It is usually the last segment of the helix, or the first segment of the loop after the helix, try making it both and rebuild/work the area.
Tryptophan Not many of these in the puzzles, but they can make or break your score. Work them for a long time, try every possible position. These can make the biggest total impact to your score.
With any of these sidechain changes, I will often make the change, if the change in the score is a big drop, I will wiggle it out until it stops, nudge it a bit, and wiggle it out again, then shake. If it is not much of a change, I will nudge it and wiggle it out 2 or 3 times before shaking. If it fails to yield additional points, you can always restore best to go back to the better score. This part of the game takes some patience, since large gains are rare, and even tiny gains can be difficult to obtain at times.
End Game Edit
Before moving on to end game it's important to go through mid game steps 2 or 3 times, typically I would not recommend doing the following steps until after the puzzle, is already the absolute best that you can make it. It is my belief that these steps should not be done until after you are finished with the puzzle and do not intend to do anything else.
- Local_Wiggle_Strategy is one of the most effective techniques of boosting score. It is also for many people one of the most effective techniques for keeping them out of the top folders because they do this technique far too early. The reason is, doing this will greatly increase the Mojo of your puzzle and often prevent your from making gains with the mid game strategies. Once I am satisfied with the results of the puzzle, I will employ this tactic, first walking the backbone from one end to the other in 4 and 5 piece increments, and then one more time in 2 and 3 piece increments. Sometimes if the final score is close, I may walk it yet again in 2 and 3 piece increments to gain a couple more points.
- It is very important after the above strategy to pull and wiggle the puzzle several times after doing so stops yielding points.
- Global_Lock/Wiggle_Strategy is another strategy which I employ after Local_Wiggle_Strategy fails to yield points, or becomes frustrating from such small yields.
- Following the above, I will start at the end of a protein, lock a segment two off from the very end, and wiggle only the end piece. After wards move the lock up one, and repeat until moving the lock about 5 times no longer yields a gain. You can do this to both ends of the protein structures. Do not lock the end piece itself.
- Finally I will lock the entire structure except the sidechains and do a global wiggle, afterwards unlocking it and doing another global wiggle. Each of them lasting about 2 full minutes. This typically yields about 1/4 of a point, but in my mind is a bit of a seal which signifies the puzzle is the best that I could accomplish within my ability and the time alloted.
After the above end game steps have been taken, it's time to move on to another puzzle or work on a group solution, as i find the endgame steps make it nearly impossible to increase your score any further once the Mojo sets in.
Good Luck & Happy Folding!