Pdb Remove Similar

Question

[Stef]

Hello everyone.

I have noticed that pdb.org has an option "remove similar". Can someone explain how it works? My assumption until now was that it compares pairwise all against all and if they have more than X% identity it removes one of the two in the pair.

Or does it remove both?

If it removes only one then which one does it choose?

Also how does it calculate pairwise %identity? With local or global alignment?

Answer 1

[Pierre Lindenbaum]

The "Help" button redirects to this page: http://www.rcsb.org/pdb/statistics/clusterStatistics.do

Algorithm for Removing Similar Sequences

The query implementation for removing similar sequences is based on pre-calculated clusters of protein chains. All protein chains of at least 20 amino acids are clustered by blastclust at 100%, 95%, 90%, 70%, 50%, 40%, and 30% sequence similarity....

(....)

[Stef]

Ah thanks a lot. But we should be careful of this:

"Sequence similarity is defined on a chain basis, but results are returned on a structure basis."

Which means that between step A.3 and A.4 we can and will end up with chains in our proteins over our threshold.

So the answer is: Chooses the highest quality protein from every cluster which was calculates from local alignments with >90% coverage of both sequences and >X% identity