25.12.2014 Views

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 ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Why can’t we get it Right Have not<br />

we tried enough<br />

• Descriptors No, we have plenty (e.g., 1000’s in<br />

Dragon)<br />

• Datamining methods No, we also have plenty (e.g.,<br />

SAS)<br />

• Training set statistics NO, it does not work<br />

• Test set statistics Maybe, but it is still insufficient<br />

So…what else can we do<br />

• Change the success criteria! Leave behind the phase of<br />

“narcissistic” <strong>model</strong>ing <strong>and</strong> focus on external<br />

predictivity <strong>and</strong> experimental validation.<br />

• Recognize <strong>QSAR</strong> as an empirical <strong>data</strong> <strong>model</strong>ing<br />

approach: just do it any (all) way you like but<br />

VALIDATE on independent <strong>data</strong>sets!

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!