Select a couple of amino acids in the binding site of your protein and retrieve, in a matter of minutes, other structures in the database which contain a similar subpocket. Similar meaning, similar features arranged similarly in 3D space. This is what the new subpocket search feature will allow you to do in the next release of 3decision! 

Why is that interesting? This allows you to identify potential off-targets or find a chemical matter that fits in a similar environment in other structures, find structures of your target that have a specific conformation of the binding site (e.g. type II binding sites in kinases, closed conformation of HSP90). 

This precious feature is a result of several years of hard work. It all started with Peter Schmidtke’s Ph.D. thesis that focused on binding site similarity calculations. Peter is 3decision’s product manager and one of the creators of fpocket - an open-source tool to identify and characterize binding sites. His work established the basis for our pocket characterization leading up to the kind of searches we are introducing today.

The ligand-binding site is decomposed into features with geometric descriptors, these features are then pairwise compared and also stored in the database. This decomposition allows us to search for similarities for the full binding sites or, like in this case, for parts of it. 

How does the subpocket search work?

• The user selects a set of amino acids in the binding site.

• A panel opens allowing to parametrize the query (minimum score and coverage of the results).

• Once results are back, the user can automatically superpose the hit to the query using the matching residues by simply clicking on a retrieved structure. 

Check out this short video to see how this works and stay tuned for future exciting release notes.

3decision’s blog has been moved!