Designing Multi‐target Compound Libraries with Gaussian Process Models

Bieler, Michael; Reutlinger, Michael; Rodrigues, Tiago; Schneider, Petra; Kriegl, Jan M.; Schneider, Gisbert

doi:10.1002/minf.201501012

Cited by 10 publications

(4 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This was likely due to the fact that the ChEMBL set was larger and more structurally diverse. 35 173 human targets with K i bioactivity data were extracted from ChEMBL and used to build Nai ̈ve Bayes, logistic regression, and random forest (RF) models using Morgan fingerprints or FP2 that were in turn used for target inference. 36 Seven-fold cross-validation and temporal cross-validation demonstrated that cutoffs that were more potent had better accuracy and Matthew's correlation coefficient (MCC).…”

Section: ■ Introductionmentioning

confidence: 99%

“…A set of G-protein coupled receptor targets for 5-HT2c, melanin-concentrating hormone, and adenosine A1 were used to build Gaussian process models with CATS2 descriptors which were more predictive for a test set than models developed from a proprietary Boehringer Ingelheim dataset. This was likely due to the fact that the ChEMBL set was larger and more structurally diverse . 173 human targets with K i bioactivity data were extracted from ChEMBL and used to build Naïve Bayes, logistic regression, and random forest (RF) models using Morgan fingerprints or FP2 that were in turn used for target inference .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Bioactivity Comparison across Multiple Machine Learning Algorithms Using over 5000 Datasets for Drug Discovery

et al. 2020

View full text Add to dashboard Cite

Machine learning methods are attracting considerable attention from the pharmaceutical industry for use in drug discovery and applications beyond. In recent studies, we and others have applied multiple machine learning algorithms and modeling metrics and, in some cases, compared molecular descriptors to build models for individual targets or properties on a relatively small scale. Several research groups have used large numbers of datasets from public databases such as ChEMBL in order to evaluate machine learning methods of interest to them. The largest of these types of studies used on the order of 1400 datasets. We have now extracted well over 5000 datasets from CHEMBL for use with the ECFP6 fingerprint and in comparison of our proprietary software Assay Central with random forest, k-nearest neighbors, support vector classification, naïve Bayesian, AdaBoosted decision trees, and deep neural networks (three layers). Model performance was assessed using an array of fivefold cross-validation metrics including area-under-the-curve, F1 score, Cohen’s kappa, and Matthews correlation coefficient. Based on ranked normalized scores for the metrics or datasets, all methods appeared comparable, while the distance from the top indicated that Assay Central and support vector classification were comparable. Unlike prior studies which have placed considerable emphasis on deep neural networks (deep learning), no advantage was seen in this case. If anything, Assay Central may have been at a slight advantage as the activity cutoff for each of the over 5000 datasets representing over 570,000 unique compounds was based on Assay Central performance, although support vector classification seems to be a strong competitor. We also applied Assay Central to perform prospective predictions for the toxicity targets PXR and hERG to further validate these models. This work appears to be the largest scale comparison of these machine learning algorithms to date. Future studies will likely evaluate additional databases, descriptors, and machine learning algorithms and further refine the methods for evaluating and comparing such models.

show abstract

Section: ■ Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Bioactivity Comparison across Multiple Machine Learning Algorithms Using over 5000 Datasets for Drug Discovery

et al. 2020

View full text Add to dashboard Cite

show abstract

“…High-dimensional numerical target labels can be simultaneously predicted , using a variety of methods popular in the physical sciences, including linear regression (LR), k -nearest neighbors (kNN), , random forests (RF), support vector machine (SVMs), , artificial neural networks (ANNs), , and Gaussian process regressors (GPRs). − Each method has strengths and weaknesses, including dealing with high dimensional or imbalanced data, whether the algorithm is intrinsically multitarget (ingesting multiple labels without variations to the pipeline), and the degree of model transparency, interpretability, or explainability.…”

Section: Methodsmentioning

confidence: 99%

Simultaneous Prediction and Optimization of Charge Transfer Properties of Graphene and Graphene Oxide Nanoflakes from Multitarget Machine Learning

Zhuang,

Fox,

Barnard

2023

J. Phys. Chem. C

View full text Add to dashboard Cite

Considerable effort is directed toward controlling the physicochemical structure of graphene and graphene oxide, but complex structure/property relationships are difficult to identify and utilize when the materials are multifunctional and the properties are correlated. In this study, we propose and demonstrate a workflow for predicting which structural features to use to tune correlated properties simultaneously. Highly accurate multitarget regressors predict the ionization potential and electron affinity of graphene and graphene oxide nanoflakes and report the most important structural features as a basis for ensemble filtering that reflects design decisions. To challenge the approach, multiobjective optimization was used to find filters that simultaneously lower the ionization potential by −0.5 eV and raise the electron affinity by 0.5 eV. We find that the diameter of graphene nanoflakes is the most useful structural feature of graphene but is superseded by the oxygen concentration and proximity to the edges in graphene oxide. Achieving our challenging design goal was not possible, but a significant and balanced shift in the properties (in the right directions) could be obtained and accompanied by improved quality and performance. This general approach could be used to predict filters and to guide experimental design to separate samples for specific applications.

show abstract

“…A set of G-protein coupled receptor targets for 5-HT2c, melanin concentrating hormone and adenosine A1 were used to build Gaussian process models with CATS2 descriptors which were more predictive for a test set than models developed from a proprietary Boehringer Ingelheim dataset. This was likely due to the fact that the ChEMBL set was larger and more structurally diverse 33 . 173 human targets with Ki bioactivity data were extracted from ChEMBL and used to build Naïve Bayes, logistic regression and random forest models using Morgan fingerprints or FP2 that were in turn used for target inference 34 .…”

Section: Introductionmentioning

confidence: 99%

A Very Large-Scale Bioactivity Comparison of Deep Learning and Multiple Machine Learning Algorithms for Drug Discovery

Lane¹,

Foil²,

Minerali³

et al. 2020

Preprint

View full text Add to dashboard Cite

Machine learning methods are attracting considerable attention from the pharmaceutical industry for use in drug discovery and applications beyond. In recent studies we have applied multiple machine learning algorithms, modeling metrics and in some cases compared molecular descriptors to build models for individual targets or properties on a relatively small scale. Several research groups have used large numbers of datasets from public databases such as ChEMBL in order to evaluate machine learning methods of interest to them. The largest of these types of studies used on the order of 1400 datasets. We have now extracted well over 5000 datasets from CHEMBL for use with the ECFP6 fingerprint and comparison of our proprietary software Assay CentralTM with random forest, k-Nearest Neighbors, support vector classification, naïve Bayesian, AdaBoosted decision trees, and deep neural networks (3 levels). Model performance <a>was</a> assessed using an array of five-fold cross-validation metrics including area-under-the-curve, F1 score, Cohen’s kappa and Matthews correlation coefficient. <a>Based on ranked normalized scores for the metrics or datasets all methods appeared comparable while the distance from the top indicated Assay CentralTM and support vector classification were comparable. </a>Unlike prior studies which have placed considerable emphasis on deep neural networks (deep learning), no advantage was seen in this case where minimal tuning was performed of any of the methods. If anything, Assay CentralTM may have been at a slight advantage as the activity cutoff for each of the over 5000 datasets representing over 570,000 unique compounds was based on Assay CentralTMperformance, but support vector classification seems to be a strong competitor. We also apply Assay CentralTM to prospective predictions for PXR and hERG to further validate these models. This work currently appears to be the largest comparison of machine learning algorithms to date. Future studies will likely evaluate additional databases, descriptors and algorithms, as well as further refining methods for evaluating and comparing models.

show abstract

Designing Multi‐target Compound Libraries with Gaussian Process Models

Cited by 10 publications

References 18 publications

Bioactivity Comparison across Multiple Machine Learning Algorithms Using over 5000 Datasets for Drug Discovery

Bioactivity Comparison across Multiple Machine Learning Algorithms Using over 5000 Datasets for Drug Discovery

Simultaneous Prediction and Optimization of Charge Transfer Properties of Graphene and Graphene Oxide Nanoflakes from Multitarget Machine Learning

A Very Large-Scale Bioactivity Comparison of Deep Learning and Multiple Machine Learning Algorithms for Drug Discovery

Contact Info

Product

Resources

About