I'm trying to find unique chains or proteins within their PDB files. Many proteins have multiple chains, but very often they are identical (say, the PDB file consists of 4 homodimers, for example, so there is one unique chain represented 8 times total). I am processing certain structural features of the chains, so processing two identical chains is redundant for my pruposes.
I am using Biopython and the way I'm currently identifying unique chains is looking through each PDB file, checking the sequences of the chains and comparing them to one another using alignments (specifically pairwise2.align.globalxx(seq1, seq2). I consider a chain non-unique if it shares >90% sequence similarity to another chain. (Most chains have 100+ residues.)
My question is, is this threshold appropriate? Should I go lower, maybe 80%, or the opposite, higher? For now, it seems to work fine, but I'll be processing a lot of PDB files and want to be on the safer side. If I accidentally process identical chains, that's not much of an issue, but it adds unnecessary time to the processing. Is there maybe a source/article to help with what sequence similarity I should be using? I've found multiple articles, but they usually talk about homology (i.e., common ancestry of proteins), but I'm only doing this within the protein itself and its subunits (chains), so this is of no use for me.
