Core Theory Edit
My method is based upon the following ideas:
- A properly folded protein will be in its lowest energy state
- Therefore, it will be very robust to being smashed around
- Proteins have a tendency to fall into "local minima"
- You need to give the protein a bit of a kick to get out of this state
- The protein is in water (or a similar solution)
Starting Out Edit
It's important to know your protein. Make friends with it. Talk to it, find out its name... okay, you probably don't need to go that far.
Look for common structure combinations:
- A beta sheet probably needs other sheets to be happy
- Beta sheets generally join up slightly curved -- a very slight spiral
- Helices join up slightly off parallel, and usually with something of similar length
- Loops at the edges of sheets / spirals can have a "big spoon, little spoon" connection with other loops
- Most importantly -- hydrophobic acids fear water and hate hydrophilic chains. Point them inside, or failing that, try to match them up with other hydrophobic acids
Mess around with the protein:
- Find places to put bits of the protein, and see if they "stick"
- Look for chains that have similar hydrophobic/hydrophilic sequences and try to get them together
- Work out how stable it is to start with -- what does it take to get the protein to blow up?
- This can influence strategy later on
The Process Edit
My strategy is basically a "smash it and see" strategy. You want to knock the protein out of the local minima, in the hope that it'll find another lower energy point. Ideally, it should be done just enough to get out of the current hole, but no more.
For unstable proteins Edit
I define an unstable protein as something that breaks apart when you sneeze at it (you did work this out previously, right?). The main aim here is to turn it into a strong protein, don't worry too much about it looking nice at this stage.
- Find the piece that is the most unstable, and see where it wants to go.
- You may need to break apart the protein in order to do this. Don't worry, just CTRL+R if you get really lost.
- Try to nudge this piece towards that place without disrupting the protein.
- If you can't do that, hunt for another piece.
- If you manage to get it to go somewhere near that place, test the protein again to find out its stability.
For stable proteins Edit
These can be tricky to work with, because they may settle very hard into a state that is not their final resting state. You need to make a judgement here. If the protein is not compact, or some structure looks a bit out of place, you'll need to get the protein into a better position. Remember what the protein looked like, then carry out a controlled smash:
- Drag part of the protein into another part, then wiggle until it just starts to break apart, drag the part back a bit, then wiggle again.
- Alternatively, set a band between two extremes of the protein, wiggle until it starts to break apart, then get rid of (or disable) the band.
- Alternatively, lock off a small section and do a rebuild -- that usually causes a fair amount of chaos
Sometimes you need to tie down most of the protein, and just smash a bit of it into another position. In this case, I've found that many bands (say ~10) pinned to their starting location works to keep the protein in check (and it can be funny to watch it struggle to try to get free).
Now, it's just a matter of finding easy points. You want the protein to shuffle round as little as possible, but not so little that it goes back to where it started. I generally stick to the following process:
- Do something to disrupt the protein slightly
- Wiggle until the score counter stops changing
- Shake, rinse, repeat
In the event of the new score being lower than the score it started out with before the shake, I make a judgement as to whether I think the protein is better despite having a lower score (if not, then revert to best / saved / etc.).
Here's some of the things I try:
- Attach a band between two areas of the protein, possibly reducing the strength if it disrupts things too much. Alternate between enable/disable for the band.
- Tweak helices / sheets a couple of steps, either rotate or straighten
- Move something somewhere it clearly doesn't want to go (i.e. hydrophobic sidechain near a hydrophilic sidechain)
- Move something away from the centre of the protein (and wiggle it back, hopefully to a new position)
- Lock off small bits and rebuild -- you're looking for rebuilds that don't change the protein much
- It can help to save first here -- the undo may not be enough to go back to a "good" version
- If you rebuild the end of a sheet or helix, the rebuild will try to twist the protein to make that sheet/helix straighter
- If you have a tricky twist in a helix/sheet, it can sometimes help to convert it to a loop, then rebuild (the rebuild is more disruptive of loops)