Dear All
According you, would it make sense to map disorder and solvent accessibility on a multiple alignment (built on a cluster of homolog proteins) in order to infer on hypotetical interaction sites?
Best regards
In my opinion it not particularly useful to have both on the same plot but rather separate out the sites into ordered and disorder first using the disorder prediction tools like IUPred.
If you think some of your interaction sites are in globular domains then plot the solvent accessibility for these. But the disordered regions are already solvent accessible so there isn't anything added by running both predictions. Most solvent accessibility scoring methods are trained on structural data which is obviously missing for disordered protein segments.
If the disorder pattern is shared by all of your homologues and they all retain the same interactors then calculating something like the relative local conservation or ConScore would be useful. The RLC is plotted for sequences in SLiMFinder, along with disorder however, it is designed to work on sequences that convergently evolve an interaction site. You could try putting your sequences through it - though if they are all related then it won't work properly.
Yes. IMHO, there is a strong biological rational to support your approach. Both are extensively studies sequence features.
Solvent accessibility can tell you whether a group of residues in your alignment share a similar trend in solvent exposure. This can give you information that whether the given residue is accessible or buried for a hypothetical interactions. See the following paper for background reading on solvent accessibility 1, and another one on application of solvent accessibility to analyze catalytic residues in enzymes 2.
Disorder regions in proteins are reported to be mediators of protein-protein interactions. See this papers and the disordered proteins wiki page for more details.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
"..the disordered regions are already solvent accessible so there isn't anything added by running both predictions." I disagree with this. Proteins are not 1-dimensional entity. If you think from a structural perspective a buried residue (solvent inaccessible) could also be disordered.
If the residue is buried then it would be constrained in 3D space. If you look in DisProt then the disordered segments are not buried. A buried residue could be flexible within the core of a globular domain but its not unconstrained in the same way a disordered region is.