QSAR in safety evaluation and risk assessment / Huixiao Hong, editor.

Format:

Book

Contributor:

Hong, Huixiao, editor.

Language:

English

Subjects (All):

Drugs--Structure-activity relationships.

Drugs.

QSAR (Biochemistry).

Quantitative Structure-Activity Relationship.

Consumer Product Safety.

Environmental Pollutants.

Toxicity Tests.

Medical Subjects:

Quantitative Structure-Activity Relationship.

Consumer Product Safety.

Environmental Pollutants.

Toxicity Tests.

Physical Description:

1 online resource (566 pages)

Edition:

1st ed.

Other Title:

Quantitative structure–activity relationship in safety evaluation and risk assessment

Place of Publication:

Kidlington, England : Matthew Deans, [2023]

Summary:

QSAR in Safety Evaluation and Risk Assessment provides comprehensive coverage on QSAR methods, tools, data sources, and models focusing on applications in products safety evaluation and chemicals risk assessment. Organized into five parts, the book covers almost all aspects of QSAR modeling and application. Topics in the book include methods of QSAR, from both scientific and regulatory viewpoints; data sources available for facilitating QSAR models development; software tools for QSAR development; and QSAR models developed for assisting safety evaluation and risk assessment. Chapter contributors are authored by a lineup of active scientists in this field. The chapters not only provide professional level technical summarizations but also cover introductory descriptions for all aspects of QSAR for safety evaluation and risk assessment.

Contents:

Front Cover

QSAR IN SAFETY EVALUATION AND RISK ASSESSMENT

Copyright

Contents

List of contributors

Preface

1 - QSAR facilitating safety evaluation and risk assessment

Introduction

Data sources for QSAR

QSAR methods

Evaluation of QSAR models

Machine learning and deep learning accelerate QSAR development

Machine learning

Deep learning

Perspectives

References

I - Methods and advances of QSAR

2 - Development of QSAR models as reliable computational tools for regulatory assessment of chemicals for acute tox ...

Introduction: growing regulatory pressure to develop alternative computational methods for chemical toxicity assessment

Comparison of computational approaches for chemical toxicity prediction

Contrasting alerts and QSAR-based predictions of acute toxicity

Integration of interpretative QSAR models and chemical alerts

The continuing importance of data quality and curation in the age of big data and AI

Biomedical knowledge mining to identify mechanistic pathways underlying chemical toxicity effects

Conclusions and perspectives

Acknowledgments

3 - Neural network-based descriptors as input for QSAR

Deep learning-based methods for generating descriptors

Element expression and summarization

Latent representation of the whole structure

Other methods for generating descriptors

Black box approach in deep learning-based descriptors

Summary

Combination with different modalities: phenotype-based descriptors

Future perspectives

Further reading

4 - Decision forest-a machine learning algorithm for QSAR modeling

Decision forest algorithm

Two-class classification

Multiclass classification.

QSAR models developed using decision forest

QSAR models for predicting estrogen receptor activity

QSAR models for predicting androgen receptor activity

QSAR models for predicting drug-induced liver injury

Conclusion remarks

5 - Integrated modeling for compound efficacy and safety assessment

Compound representation

Molecular representation

MOA representation

Datasets for compound discovery

Virtual screening

Quantitative structure-activity relationship

Generative models

Read-across

Biomarker discovery

Systems pharmacology

Knowledge graph-based approaches for chemical safety and drug design

Conclusions

6 - Deep learning quantitative structure-activity relationship methods for chemical toxicity prediction and risk as ...

Deep learning methods

Deep neural network

Convolutional neural network

Recurrent neural network

Graph neural network

Key DL techniques for QSAR researches

Hyperparameter optimization and overfitting controlling

Model visualization and interpretation

Recent advances in DL-based QSAR researches in toxicity prediction and risk assessment

Consensus, heterogenous, and other DL methods

Free available DL-based tools for chemical toxicity prediction

Conclusions and future perspectives

7 - Predictive modeling approaches for the risk assessment of persistent organic pollutants (POPs): from QSAR to ma ...

Significant breakthroughs in QSAR modeling of POPs

Current advancements and guidelines for QSAR model development of POPs

Different molecular endpoints for the classification of POPs.

Molecular descriptors utilized in the QSAR modeling of POPs

Statistical and ML-based approaches for model development of POPs

Classical approaches for QSAR model development

ML-based QSAR approaches

Contemporary QSAR tools for PBT analysis of POPs

8 - Machine learning-based QSAR for safety evaluation of environmental chemicals

ML-driven QSAR modeling

Molecular structure descriptors

ML-driven fingerprints

Algorithms

ML-driven QSAR applications

High-throughput screening toxicity data-driven QSAR model

Read-across coupled with ML-based QSAR

Adverse outcome pathway-based modeling

Multiomics-based modeling

Challenges in QSAR model construction

9 - Advances in QSAR through artificial intelligence and machine learning methods

Roadmap of QSAR method

QSAR methods based on molecular descriptors

Artificial intelligence in drug screening

Iterative integration of different models in QSAR

Neural network learning

Artificial neural network-quantitative structure-activity relationship

Machine learning models-Quantitative structure-activity relationship

Deep learning-Quantitative structure-activity relationship

Decision tree algorithms

Random forest

Supervised learning

Intrinsic proximity measure

AdaBoost classifier

Partial least squares regression

Software's available for QSAR

Concluding remarks

10 - Advances of the QSAR approach as an alternative strategy in the environmental risk assessment

The principal aspects of ERA

New approach methodologies (NAMs)

The QSAR approach and its fundaments

QSAR theory

Similarity influence in the QSAR models

Aims of QSAR modeling.

General methodologies of the QSAR models

Structure-activity relationship approach

Multivariable linear regression

Comparative molecular fields analysis

QSAR modeling

Training

Modeling processes

Validation and interpretation of the QSAR models

Features, contributions, and advances of QSAR modeling in ERA processes

Future perspective of QSAR modeling within ERA approach

11 - QSAR modeling based on graph neural networks

QSAR models for the management of chemicals

QSAR models for predicting chemical properties

QSAR models for screening PBT chemicals

GNN algorithm

Basic principles of GNNs

Principles of graph attention networks

QSAR model based on graph attention network

GNN for QSAR modeling

GNN models for chemical property predictions

GAT models for screening PBT chemicals

Applicability domains for GNN-based QSAR models

II - Tools and data sources for QSAR

12 - Modeling safety and risk assessment with VEGAHUB

The global needs of modern society about risk assessment and safety

The VEGAHUB components

The VEGA in silico models

The read-across tools

The VERMEER tools

The ToxEraser tool

The JANUS tool

The architecture and the conceptual basis within VEGAHUB

The use of VEGAHUB for safety and risk assessment

The role of VEGAHUB within a larger network

13 - Recent advancements in QSAR and machine learning approaches for risk assessment of organic chemicals

Brief overview of the methodologies used for QSAR modeling in predictive toxicology

Data collection and data preparation

Molecular descriptor computation

Feature selection

Modeling algorithms

Basics of ML

ML algorithms

SVM.

Decision tree

k-NN

RF

Naive Bayes

Gaussian process regression

Neural networks

QSAR model validation

Other methods

Read-across structure-activity relationship

QSPR applications in toxicity prediction of organic chemicals

Prediction of endocrine disruptive chemical toxicity using QSTR

Prediction of miscellaneous chemical toxicity using QSTR

Conclusion

Acknowledgment

14 - admetSAR-A valuable tool for assisting safety evaluation

Basic architecture of admetSAR

Details of admetSAR

Data collection and preparation

Data representation and model building

Model evaluation and prediction

Property optimization module-ADMETopt

Usage of admetSAR

Submission web page

Output web page

ADMETopt web page

Applications of admetSAR

Evaluation of newly synthesized molecules

ADMET evaluation in virtual screening

ADMET evaluation in mechanism exploration

Comparison with other tools

Conclusions and outlook

15 - QSAR tools for toxicity prediction in risk assessment-Comparative analysis

The basic information of toxicity prediction software package

Development background

Toxicity prediction endpoints

User experience

Modeling methods of the toxicity prediction software packages

Type of endpoints

Raw datasets

Structural characterization method

Application domain

16 - Fast and efficient implementation of computational toxicology solutions using the FlexFilters platform

The "filter" concept

Syntax for the filter calls

Frequently used filters in FlexFilters platform

Building FlexFilters modules

Applying the modules for prediction.

Examples of computational toxicology solutions built using the FlexFilters methodology.

Notes:

Includes bibliographical references and index.

Description based on print version record.

Other Format:

Print version: Hong, Huixiao QSAR in Safety Evaluation and Risk Assessment

ISBN:

9780443153402

044315340X