Skip to main content

Table 1 Prediction sensitivity (%) of the top three predictions by different geometric approaches and their comparison to IBIS.

From: Knowledge-based annotation of small molecule binding sites in proteins

Method* Top1 Top2 Top3
IBIS100 73 89 89
IBIS90 75 91 91
IBIS80 72 88 88
LIGSITEcs 71 79 85
PASS 58 67 75
Q-SiteFinder 52 60 75
SURFNET 42 58 62
  1. *IBIS100, IBIS90, IBIS80 dataset contains unbound query proteins for which the average sequence identity between the unbound protein and members of the binding site clusters containing the bound homolog is no more than 100%, 90% and 80% respectively.