Table 1.
Prediction of ligands by ProBiS-ligands on 500 test proteins
Sequence identity cutoff of template proteinsa | ||||
---|---|---|---|---|
30% | 20% | 10% | ||
Models | Ligand similarityb | 0.55 | 0.34 | 0.32 |
MCC | 0.41 | 0.13 | 0.09 | |
Precision | 0.42 | 0.17 | 0.14 | |
Recall | 0.45 | 0.18 | 0.15 | |
Experimental | Ligand similarityb | 0.61 | 0.46 | 0.40 |
MCC | 0.54 | 0.33 | 0.28 | |
Precision | 0.56 | 0.38 | 0.33 | |
Recall | 0.57 | 0.36 | 0.31 |
aWe excluded from the template libraries all protein structures with sequence identity >30%, >20% and >10% to the corresponding query proteins.
bExpressed with Tanimoto coefficient.