Core Exists filter

Started by bkoep

bkoep Staff Lv 1

There have been several design puzzles now with the Core Exists filter, but we've never really explained how it works or why it's there. This post is meant to explain a bit about the filter and how you can use it to improve your score in design puzzles.

As you know, a protein is made up of a long chain of amino acids, but the chain folds up into a globular, compact shape. One of the main forces that drives folding is the burial of hydrophobic residues. Since the protein is surrounded by water, the only way to hide the hydrophobic residues is to fold them into the center of the protein and shield them from water with polar residues. This is why the most stable proteins have well-packed cores with lots of buried hydrophobics.

When designing a protein, we want to make sure that it has a well-packed core of hydrophobic residues to stabilize the fold. The "Core Exists" filter calculates the solvent-accessible surface area of each residue to determine which residues are buried and which residues are exposed to water. Then it checks that the number of buried residues meets some threshold (usually 30% of the total number of residues). Additional buried residues get a score bonus, but any fewer than the threshold and your protein will get a penalty. If a protein does not pass the filter, the Show checkbox will highlight residues that are not buried—you should try to pack more of these into the core of your protein.

Note that, in puzzles with multiple chains, residues at the interface between chains are counted as buried. So, while individual chains should have their own cores, large well-packed interfaces will also contribute to the filter score.

Hope this helps—we've been seeing some great designs from these puzzles! If you have a question, please leave a comment below.

karstenw Lv 1

I think most of my multi chains rely heavily on the interface for burial. Should I be concerned If I don't see much of a visual core in the individual chain; or should I assume that as long as my core is scoring well, that the interfaced burials are adequate?

bkoep Staff Lv 1

You should strive for designs that have individual cores.

There are cases in nature in which the subunits of a multimer do not have individual cores (tropomyosin in muscle, for example, is a dimer that consists simply of two helices wound together). Such structures can be stable, and may even get you a lot of points in a design puzzle. But these kinds of designs don't exploit the full potential of protein interface design, and they're not very interesting.

At the Baker lab, we get really excited about designs that have individual cores as well as strong interfaces. These structures receive lots of attention and are more likely to enter our pipeline of analysis and testing. A good example of this would be MurloW's scientist-shared solution from Puzzle 691, which was used as the starting point for Puzzle 725.

AsDawnBreaks Lv 1

I don't hand mutate, I just let the mutate all do the work for me (unless there is a particularly bad scoring sidechain, then I have a script that tries all possibilities for the best sidechain). Is mutate all configured to look for these, and am I doing the "wrong" thing by just using that?

bkoep Staff Lv 1

For the most part, mutating a residue will not affect how that residue contributes to the Core Exists filter. The residue will remain as "core" or "surface" and your Core Exists status will not change.

However, there are occasional cases where a mutation drastically changes the solvent-accessible surface area of the residue, and the mutation will affect the filter. Mutate All is not aware of the filter, and it will sometimes make mutations that cause the filter to fail. If you're having trouble meeting the Core Exists filter requirement, you could try hand-mutating residues that look like they are in the core but are still highlighted by the filter.

Note that the Core Exists filter does not care whether your core residues are hydrophobic or polar. You can have a core full of polar residues and the filter will still pass, although your overall score will drop. Once you have a core, Mutate All is pretty good about mutating those residues to hydrophobics.

Bruno Kestemont Lv 1

If I understand well, the distance between the backbones of the residues (not the distance between sidechains) is calculated to make the filter.

Does it imply that I'm "sicked" to short hydrophobic residues inside?

However, I have the impression that a full "inside" of long hydrophobics would also be good against water. No?