Classifying the world anti-doping agency’s 2005 prohibited list using the chemistry development kit fingerprint

  • Authors:

  • Edward O. Cannon;John B. O. Mitchell

  • Affiliations:

  • Unilever Centre for Molecular Science Informatics, Department of Chemistry, University of Cambridge, Cambridge, United Kingdom;Unilever Centre for Molecular Science Informatics, Department of Chemistry, University of Cambridge, Cambridge, United Kingdom

  • Venue:
  • CompLife'06 Proceedings of the Second international conference on Computational Life Sciences
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

We used the freely available Chemistry Development Kit (CDK) fingerprint to classify 5235 representative molecules taken from ten banned classes in the 2005 World Anti-Doping Agency’s (WADA) prohibited list, including molecules taken from the corresponding activity classes in the MDL Drug Data Report (MDDR). We used both Random Forest and k-Nearest Neighbours (kNN) algorithms to generate classifiers. The kNN classifiers withk = 1 gave a very slightly better Matthews Correlation Coefficient than the Random Forest classifiers; the latter, however, predicted fewer false positives. The performance of kNN classifiers tended to decline with increasing k. The performance of the CDK fingerprint is essentially equivalent to that of Unity 2D. Our results suggest that it will be possible to use freely available chemoinformatics tools to aid the fight against drugs in sport, while minimising the risk of wrongfully penalising innocent athletes.