Best practices for chemical data curation and QSAR model ...
Best practices for chemical data curation and QSAR model ...
Best practices for chemical data curation and QSAR model ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Activity r<strong>and</strong>omization: <strong>model</strong> robustness<br />
Struc.1<br />
Pro.1<br />
Struc.2<br />
Pro.2<br />
0.7<br />
Struc.3<br />
.<br />
.<br />
Struc.n<br />
Pro.3<br />
.<br />
.<br />
Pro.n<br />
q2<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
The lowest q 2 = 0.51 in the top 10 <strong>model</strong>s<br />
The highest q 2 =0.14 <strong>for</strong> r<strong>and</strong>omized <strong>data</strong>sets<br />
0.1<br />
Struc.1<br />
Struc.2<br />
Struc.3<br />
.<br />
.<br />
Struc.n<br />
Pro.1<br />
Pro.2<br />
Pro.3<br />
.<br />
.<br />
Pro.n<br />
0<br />
-0.1<br />
0 10 20 30 40 50 60 70<br />
Number of Variables<br />
Training set with real property values is<br />
expected to produce much higher q 2 values<br />
than the same set with r<strong>and</strong>omized<br />
property values.