Table 1. Databases of protein surface patterns.

Database	No. of surfaces	Time (sec.)
CASTp	2 000 000+	43
PDBSelect (90%)	241 119	12
PDBSelect (25%)	55 840	2

The full CASTp database contains all computed pocket and void surfaces in the Protein Data Bank. The PDBSelect databases contain the subset of surface patterns found on protein structures contained in the corresponding PDBSelect list at different levels of sequence identity (8). The column ‘Time’ refers to the number of seconds it takes to perform the search using pocket 19 of protein myoglobin from Physeter catodon (PDB id = 109 m) as the database query (actual times will vary depending on network traffic).