I have a number of PDBs which have at least 85% homology. What would be the best by way to identify the structure which best represents all the PDBs.
I tried to use Chimera's ensemble cluster but my PDBs have an unequal amount of atoms and cannot be achieved through this route.
How about going by structure quality, there are a few metrics so far I remember, resolution being one of them.
For me, it depends on the objective of your project. 1. select the one with higher resolution 2. if presence/absence of ligand matters, consider it while selecting. 3. Mutation of certain amino acids 4. Orthorhombic or monoclinic? 5. make qualitative check (PROCHECK, for example) and select the best one.
As you want to pick the most representative one, make a sequence alignment to have a bird's eye view of the set. You can see some sequences, that you may not want to consider.
Have you tried to use SWISS-MODEL Repository ?
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
A structure that is least "furthest away" from all the others
what do you mean by representative structure?
For example if I have structres A,B and C in my dataset. If i say i calcaulted a "divergence distance" between A and C to be 3 and between A and B to be 1.5. The divergence distance between B and C is also 1.5. PDB B would then become the representative structre because it most like the other structres A and C. The term "divergence distance" and the number values were completely made up. I used them to hopefully clarify my question.