Visual Interpretation of Kernel‐Based Prediction Models
- 5 September 2011
- journal article
- research article
- Published by Wiley in Molecular Informatics
- Vol. 30 (9) , 817-826
Abstract
Statistical models are frequently used to estimate molecular properties, e.g., to establish quantitative structure‐activity and structure‐property relationships. For such models, interpretability, knowledge of the domain of applicability, and an estimate of confidence in the predictions are essential. We develop and validate a method for the interpretation of kernel‐based prediction models. As a consequence of interpretability, the method helps to assess the domain of applicability of a model, to judge the reliability of a prediction, and to determine relevant molecular features. Increased interpretability also facilitates the acceptance of such models. Our method is based on visualization: For each prediction, the most contributing training samples are computed and visualized. We quantitatively show the effectiveness of our approach by conducting a questionnaire study with 71 participants, resulting in significant improvements of the participants’ ability to distinguish between correct and incorrect predictions of a Gaussian process model for Ames mutagenicity.Keywords
This publication has 21 references indexed in Scilit:
- StructRank: A New Approach for Ligand-Based Virtual ScreeningJournal of Chemical Information and Modeling, 2010
- Applications of machine learning and computational intelligence to drug discovery and developmentDrug Development Research, 2010
- Gaussian Processes for Classification: QSAR Modeling of ADMET and Target ActivityJournal of Chemical Information and Modeling, 2010
- Enhanced SAR Maps: Expanding the Data Rendering Capabilities of a Popular Medicinal Chemistry ToolJournal of Chemical Information and Modeling, 2009
- Benchmark Data Set for in Silico Prediction of Ames MutagenicityJournal of Chemical Information and Modeling, 2009
- A Probabilistic Approach to Classifying Metabolic StabilityJournal of Chemical Information and Modeling, 2008
- Predicting Lipophilicity of Drug‐Discovery Molecules using Gaussian Process ModelsChemMedChem, 2007
- Similarity to Molecules in the Training Set Is a Good Discriminator for Prediction Accuracy in QSARJournal of Chemical Information and Computer Sciences, 2004
- SVM-Based Feature Selection for Characterization of Focused Compound CollectionsJournal of Chemical Information and Computer Sciences, 2004
- MUTAGENICITY AND ANTIMUTAGENICITY OF FLAVONOIDS EXTRACTED FROM MILLINGTONIA HORTENSIS L.The Journal of Toxicological Sciences, 1998