Skip to main content

Table 3 Prediction performance for selected MUV dataset (F1-score)

From: End-to-end learning for compound activity prediction based on binding pocket information

MUV ID

Pocket structure (Proposed)

Sequence (Tsubaki et al.)

Docking (AutoDock Vina)

600

0.138

0.029

0.031

692

0.031

0.026

0.014

859

0.041

0.019

0.023

852

0.241

0.199

0.011

548

0.329

0.263

0.020

832

0.122

0.131

0.041

689

0.011

0.021

0.051

846

0.112

0.112

0.009

466

0.010

0.006

0.117

Average

0.115

0.090

0.035