Filtered circular fingerprints improve either prediction or runtime performance while retaining interpretability

Abstract

BackgroundEven though circular fingerprints have been first introduced more than 50 years ago, they are still widely used for building highly predictive, state-of-the-art (Q)SAR models. Historically, these structural fragments were designed to search large molecular databases. Hence, to derive a compact representation, circular fingerprint fragments are often folded to comparatively short bit-strings. However, folding fingerprints introduces bit collisions, and therefore adds noise to the encoded structural information and removes its interpretability. Both representations, folded as well as unprocessed fingerprints, are often used for (Q)SAR modeling.ResultsWe show that it can be preferable to build (Q)SAR models with circular fingerprint fragments that have been filtered by supervised feature selection, instead of applying folded or all fragments. Compared to folded fingerprints, filtered fingerprints significantly increase predictive performance and remain unambiguous and interpretable. Compared to unprocessed fingerprints, filtered fingerprints reduce the computational effort and are a more compact and less redundant feature representation. Depending on the selected learning algorithm filtering yields about equally predictive (Q)SAR models. We demonstrate the suitability of filtered fingerprints for (Q)SAR modeling by presenting our freely available web service Collision-free Filtered Circular Fingerprints that provides rationales for predictions by highlighting important structural features in the query compound (see http://coffer.informatik.uni-mainz.de).ConclusionsCircular fingerprints are potent structural features that yield highly predictive models and encode interpretable structural information. However, to not lose interpretability, circular fingerprints should not be folded when building prediction models. Our experiments show that filtering is a suitable option to reduce the high computational effort when working with all fingerprint fragments. Additionally, our experiments suggest that the area under precision recall curve is a more sensible statistic for validating (Q)SAR models for virtual screening than the area under ROC or other measures for early recognition. Graphical Abstract Electronic supplementary materialThe online version of this article (doi:10.1186/s13321-016-0173-z) contains supplementary material, which is available to authorized users.

Should We Embed in Chemistry? A Comparison of Unsupervised Transfer Learning with PCA, UMAP, and VAE on Molecular Fingerprints

“…For the predictive tasks, Morgan fingerprints (FPR) were calculated for the 8314 structures by means of the RDKit library [ 44 ]. Owing to the possibility of colliding bits in fingerprints [ 45 , 46 ], we set the fingerprint vector length to 5120 bits and the radius to 2. In order to foster reproducibility, we made the scripts that are used for data preprocessing and feature engineering available already in our recent work [ 16 ].…”

Applications of machine learning in drug discovery and development

“…Feature selection before model building can improve ML models, as shown in a study by Kramer and Gutlein 51 . They were also able to detect improvements in random forest models against other ML methods such as SVMs and naive Bayes, with faster performance and fewer features used while training models.…”

Machine learning on drug-specific data to predict small molecule teratogenicity

“…Of the t-SNE combinations we attempted on our structure and teratology encodings, the t-SNE plot generated with 1,024-dimensional Morgan fingerprints 56 and a binary classification of teratological risk showed the strongest clustering relationships. Clusters were identified by visual inspection, with each point within a cluster representing a drug.…”