A total open question. How would one go about linking a function to a protein given only its structure?
I understand this is a million-dollar question, so vagueness is appreciated.
I have a predicted structure of a protein from de-novo/threading (I-TASSER) followed by extensive simulated annealing MD simulations. There is only DNA work (sequencing), no actual protein experimental work - nothing at all, this is something knew we found. There are also no template structures, i.e., homology modelling efforts have been awful.
Some thoughts I've had:
Are there di-cysteine bonds? Typically extra-cellular proteins have these over intra-cellular proteins.
What is the secondary structure content? i.e., helical vs beta sheet with regards to transmembrane proteins
What is the tertiary structure? Globular, fibrillar, etc.
Are their stretches of exposed hydrophobic residues? i.e., integral membrane proteins.
Are there are metal binding sites e.g., 3x HIS 1 ASP which is very typical for zinc?
Stretches or grooves for protein-protein interaction and/or potential hydrogen bonding?
Given that I have a structure, is there any way of comparing this to 3D structures in the PDB databank? Having said that, if there was, and a match was found, you could safely assume the homology modelling would have worked to begin with.
I appreciate any thoughts you have.
Thanks