Figure 5 is a visual representation of atom-based sensitivities for a structure in the CDK2 data set using the 5-factor KPLS model. Red indicates a tendency to increase predicted activity, whereas blue indicates a tendency to decrease it. Hovering over an atom, the N‑methyl carbon in this case, displays its numeric sensitivity value. Canvas allows in-place editing of the structure, which is followed by instantaneous updates to the predicted activity and atomic sensitivities. Since the N‑methyl carbon is clearly exerting a detrimental effect on activity, it is natural to remove it to see if predicted activity improves. Doing so results in a 10-fold increase in predicted potency, suggesting that this structural modification would be highly beneficial.
Figure 5: Visual representation of each atom’s impact on predicted activity. Red circles indicate an increase in activity, whereas blue circles indicate a decrease. By removing the detrimental N-methyl group from structure (a), a 10-fold increase in predicted potency is observed for structure (b).
Other Breakthrough Technologies
In addition to the methods presented here, the Suite 2012 release of Canvas offers versatile new tools for elimination of redundant properties via objective feature selection, detection of Bemis-Murcko scaffolds, and identification of significant pharmacophoric preferences in a lead series.
For more information on these and other Canvas features, please visit the Canvas product page. To obtain a trial of Canvas, visit our Trials page.
References
- An, Y.; Sherman, W.; Dixon, S. L. Hole Filling and Library Optimization: Application to Commercially Available Fragment Libraries. Bioorg. Med. Chem. 2012, 20, 5379-5387.
- ASINEX Platinum Collection, ASINEX Corporation, Winston-Salem, NC, USA; http://www.asinex.com/).
- Bemis, G. W.; Murcko, M. A. The Properties of Known Drugs. 1. Molecular Frameworks. J. Med. Chem. 1996, 39, 2887-2893.
- Kohonen, T. In Self-Organizing Maps. Springer Series in Information Sciences; Springer: Heidelberg, 1997; Vol. 30.
- Sastry, M.; Lowrie, J. F.; Dixon, S. L.; Sherman, W. Large-Scale Systematic Analysis of 2D Fingerprint Methods and Parameters to Improve Virtual Screening Enrichments. J. Chem. Inf. Model. 2010, 50, 771-784.
- Rosipal, R; Trejo, L. J. Kernel Partial Least Squares Regression in Reproducing Kernel Hilbert Space. Journal of Machine Learning Research 2001, 2, 97-123.
- Vapnik, V.; Cortes, C. Support-Vector Networks. Machine Learning 1995, 20, 273-297.