Skip to main content

Table 4 Prospective Performance Comparison

From: Industry-scale application and evaluation of deep learning for drug target prediction

 

Null Hyp.: DNN AUC < row AUC

 

Wilcoxon test

Sign test

Algorithm

AstraZeneca

Janssen

AstraZeneca

Janssen

XGB

7.17e−15

2.62e−15

1.27e−14

6.84e−14

MF

4.01e−09

3.23e−11

2.98e−08

3.04e−11

  1. p-values of comparing the respective algorithm prospective ROC-AUC evaluation performance in the table row with that of DNNs using two different statistical tests