1. Data mining and machine learning techniques for the identification of mutagenicity inducing substructures and structure activity relationships of noncongeneric compounds
- Author
-
Luc De Raedt, Tobias Cramer, Christoph Helma, Stefan Kramer, Helma, Christoph, Cramer, Tobia, Kramer, Stefan, and De Raedt, Luc
- Subjects
Databases, Factual ,Computer science ,Information System ,system ,computer.software_genre ,Machine learning ,automated structure evaluation ,Structure-Activity Relationship ,Artificial Intelligence ,Computational Theory and Mathematic ,Mutagen ,Structure (mathematical logic) ,Interpretation (logic) ,business.industry ,Mutagenicity Tests ,Chemistry (all) ,Computer Science Applications1707 Computer Vision and Pattern Recognition ,General Chemistry ,Computer Science Applications ,Algorithm ,Support vector machine ,Identification (information) ,artificial-intelligence ,Mutagenicity Test ,Computational Theory and Mathematics ,Ab-initio Calculations ,Artificial intelligence ,Data mining ,program ,business ,computer ,carcinogenesis ,Algorithms ,Information Systems ,Mutagens - Abstract
This paper explores the utility of data mining and machine learning algorithms for the induction of mutagenicity structure-activity relationships (SARs) from noncongeneric data sets. We compare (i) a newly developed algorithm (MOLFEA) for the generation of descriptors (molecular fragments) for noncongeneric compounds with traditional SAR approaches (molecular properties) and (ii) different machine learning algorithms for the induction of SARs from these descriptors. In addition we investigate the optimal parameter settings for these programs and give an exemplary interpretation of the derived models. The predictive accuracies of models using MOLFEA derived descriptors is similar to10- 15 %age points higher than those using molecular properties alone. Using both types of descriptors together does not improve the derived models. From the applied machine learning techniques the rule learner PART and support vector machines gave the best results, although the differences between the learning algorithms are only marginal. We were able to achieve predictive accuracies up to 78% for 10-fold cross-validation. The resulting models are relatively easy to interpret and usable for predictive as well as for explanatory purposes. ispartof: Journal of chemical information and computer sciences vol:44 issue:4 pages:1402-1411 ispartof: location:United States status: published
- Published
- 2004