# How to characterize surface features in proteins?

When looking at the structure of protein domains I notice many of them show a cleft-like structure in the same location (when aligned to each other). What is a recommended way of characterizing these clefts? Maybe some curvature measure? I want then to relate them with electrostatic potential so that I can correlate these clefts with some charge or electrostatic potential. Here I'm attaching an example of what I mean:



• It looks like a rhino beetle horn Jun 19 '20 at 20:14
• I just found this review of some methods to find pockets for interaction with ligands. It somehows explain methods that can be used to characterize what I need but it is a bit old: onlinelibrary.wiley.com/doi/full/10.1002/jmr.984 . That led me to nature.com/articles/s41401-019-0228-6 ... seem like papers worth studying, maybe after that I can answer my own question? :) Jun 19 '20 at 20:45
• @IvanP You are very welcome to answer your own question, as we currently don't have very many users, and many of us don't know much about the modeling of proteins. We are also trying to reach 90% answered here. The papers you mention in your comment, might even be helpful to be in the actual question (it's up to you). Jun 19 '20 at 21:51

Normally, if various proteins have the "same" cavities/clefts, this means that they are part of the same family and the amino acids that form the cavity are conserved.

I really don't think that the electronic properties (charges and electrostatic potential surface) are directly related to the geometrical shape. Instead, they are related to the residues that are in the cavities, the protein protonation state, etc.

Normally, the interest in protein cavities/clefs is related to their activity. So, we look for small molecules that bind to proteins in their cavities and change the protein state.

Servers like CavityPlus, for example, are focused on cavity detection and functional analysis whereas servers like CASTp, are specialized in topological analysis of protein cavities.

• Yes, they are the domains of the same protein. And I agree, charge is not related to geometrical shape, but what's interesting is that for this case it may be, or at least I want to check if it's the case. That's why I'm looking for tools that can help with this. Thanks for the answer! Jun 23 '20 at 18:19
• You have another issue with the charges: they depend on how they are calculated (they are associated with each forcefields -easy to work with- or you have to calculated -hard work due to the system size-)
– Camps
Jun 23 '20 at 19:34
• Yes. But we can consider that the charge problem is already solved. In the image there blue is positive and red is negative electrostatic potential. That was solved using a Poisson-Boltzmann solver. I also have the charge distribution using another tool. Jun 24 '20 at 4:36