Neural network activation similarity: a new measure to assist decision making in chemical toxicology

Allen, Timothy E. H.; Wedlake, Andrew J.; Gelžinytė, Elena; Gong, Charles; Goodman, Jonathan M.; Gutsell, Steve; Russell, Paul

doi:10.1039/d0sc01637c

Cited by 17 publications

(22 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the case of feed-forward or convolutional architectures a natural choice is to define a fingerprint vector for each data point that consists of the neural networks layer outputs (activations) concatenated together. This similarity measure has been shown to be useful for judging the reliability of toxicity models predictions by comparing molecules not in the training set 36 . In the case of the Molecular Transformer, which has an encoder-decoder architecture the output of the encoder layers can be used as a basis for comparing data points.…”

Section: Methodsmentioning

confidence: 99%

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

2021

View full text Add to dashboard Cite

Organic synthesis remains a major challenge in drug discovery. Although a plethora of machine learning models have been proposed as solutions in the literature, they suffer from being opaque black-boxes. It is neither clear if the models are making correct predictions because they inferred the salient chemistry, nor is it clear which training data they are relying on to reach a prediction. This opaqueness hinders both model developers and users. In this paper, we quantitatively interpret the Molecular Transformer, the state-of-the-art model for reaction prediction. We develop a framework to attribute predicted reaction outcomes both to specific parts of reactants, and to reactions in the training set. Furthermore, we demonstrate how to retrieve evidence for predicted reaction outcomes, and understand counterintuitive predictions by scrutinising the data. Additionally, we identify Clever Hans predictions where the correct prediction is reached for the wrong reason due to dataset bias. We present a new debiased dataset that provides a more realistic assessment of model performance, which we propose as the new standard benchmark for comparing reaction prediction models.

show abstract

Section: Methodsmentioning

confidence: 99%

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

2021

View full text Add to dashboard Cite

show abstract

“…Similar success was found by Fabian et al [14] in which a transformer rather than a Graph Neural Network was used. Other groups have used the neural fingerprints as a tool to map reaction [15] or as an extra validation step in their prediction [16] . The strategy can also be adapted to encode molecules in a way that is more relevant to natural products and therefore lead to better results in virtual screening.…”

Section: Introductionmentioning

confidence: 99%

Natural product scores and fingerprints extracted from artificial neural networks

Menke

Massa

Koch

2021

Computational and Structural Biotechnology Journal

View full text Add to dashboard Cite

show abstract

“…Similar success was found by Fabian et al [14] in which a transformer rather than a Graph Neural Network was used. Other groups have used the neural fingerprints as a tool to map reaction [15] or as an extra validation step in their prediction [16]. The strategy can also be adapted to encode molecules in a way that is more relevant to natural products and therefore lead to better results in virtual screening.…”

Section: Introductionmentioning

confidence: 99%

Natural Product Scores and Fingerprints Extracted from Artificial Neural Networks.

Menke¹,

Massa²,

Koch

2021

Preprint

View full text Add to dashboard Cite

<div>Due to its desirable properties, natural products are an important ligand class for medicinal chemists. However, due to their structural distinctiveness, traditional cheminformatic approaches, like ligand-based virtual screening, often perform worse for natural products. Based on our recent work, we evaluated the ability of neural networks to generate fingerprints more appropriate for the use with natural products. A manually curated dataset of natural products and synthetic decoys was used to train a multi-layer perceptron network and an autoencoder-like network. An in-depth analysis showed that the extracted natural product specific neural fingerprints outperforms traditional as well as natural product specific fingerprints on three datasets. Further, we explore how the activation from the output layer of a network can work as a novel natural product likeness score. Overall two natural product specific datasets were generated, which are publicly available together with the code to create the fingerprints and the novel natural product likeness score.<br></div>

show abstract

Neural network activation similarity: a new measure to assist decision making in chemical toxicology

Abstract:
Deep learning neural networks, constructed for the prediction of chemical binding at 79 pharmacologically important human biological targets, show extremely high performance on test data (accuracy 92.2 ± 4.2%, MCC 0.814 ± 0.093, ROC-AUC 0.96 ± 0.04).

Cited by 17 publications

References 48 publications

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

Natural product scores and fingerprints extracted from artificial neural networks

Natural Product Scores and Fingerprints Extracted from Artificial Neural Networks.

Contact Info

Product

Resources

About

Neural network activation similarity: a new measure to assist decision making in chemical toxicology

Abstract: Deep learning neural networks, constructed for the prediction of chemical binding at 79 pharmacologically important human biological targets, show extremely high performance on test data (accuracy 92.2 ± 4.2%, MCC 0.814 ± 0.093, ROC-AUC 0.96 ± 0.04).

Cited by 17 publications

References 48 publications

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

Natural product scores and fingerprints extracted from artificial neural networks

Natural Product Scores and Fingerprints Extracted from Artificial Neural Networks.

Contact Info

Product

Resources

About

Abstract:
Deep learning neural networks, constructed for the prediction of chemical binding at 79 pharmacologically important human biological targets, show extremely high performance on test data (accuracy 92.2 ± 4.2%, MCC 0.814 ± 0.093, ROC-AUC 0.96 ± 0.04).