Deep Learning in Genetics and Genomics

Volume 2: Advanced Applications

1st Edition - November 28, 2024
Imprint: Academic Press
Editor: Khalid Raza
Language: English
Paperback ISBN:
9 7 8 - 0 - 4 4 3 - 2 7 5 2 3 - 4
eBook ISBN:
9 7 8 - 0 - 4 4 3 - 2 7 5 2 4 - 1

Deep Learning in Genetics and Genomics: Vol. 2 (Advanced Applications) delves into the Deep Learning methods and their applications in various fields of studies, including geneti… Read more

Purchase options

World Book Day Celebration!

Save up to 30% on books and eBooks!

Shop now

Institutional subscription on ScienceDirect

Request a sales quote

Deep Learning in Genetics and Genomics: Vol. 2 (Advanced Applications) delves into the Deep Learning methods and their applications in various fields of studies, including genetics and genomics, bioinformatics, health informatics and medical informatics generating the momentum of today's developments in the field. In 25 chapters this title covers advanced applications in the field which includes deep learning in predictive medicines), analysis of genetic and clinical features, transcriptomics and gene expression patterns analysis, clinical decision support in genetic diagnostics, deep learning in personalised genomics and gene editing, and understanding genetic discoveries through Explainable AI. Further, it also covers various deep learning-based case studies, making this book a unique resource for wider, deeper, and in-depth coverage of recent advancement in deep learning based approaches. This volume is not only a valuable resource for health educators, clinicians, and healthcare professionals but also to graduate students of genetics, genomics, biology, biostatistics, biomedical sciences, bioinformatics, and interdisciplinary sciences.

Title of Book
Cover image
Title page
Table of Contents
Copyright
Dedication
Contributors
About the editor
Preface
Acknowledgments
About the book
Chapter 1. Deep learning in predictive medicine exemplified by AI-mediated flu surveillance in USA
1 Introduction
2 Hypotheses
2.1 Research assumptions
3 Methodology
3.1 The research focus
3.2 US population data
3.3 US influenza vaccine coverage
3.4 US influenza epidemics
3.5 Multicollinearity, R, and T-tests
3.5.1 Interpretation of VIF and TI
3.6 Data analysis
4 Results
4.1 Prototype model for US influenza vaccine coverage
4.1.1 Model name
4.1.2 Model structural equation
4.1.3 Model statistics
4.1.4 Multicollinearity test
4.1.5 Comparative analysis
4.1.6 Projected influenza vaccine coverage for the US population
4.2 Estimation of influenza transmission rate in America
4.2.1 Fitted trend equation for influenza transmission rate in USA
4.2.2 Model statistics
4.2.3 Comparative analysis
4.2.4 Trend analysis for projected influenza transmission rate in the US
4.3 Estimation of seasonal influenza outbreak in America
4.3.1 Model name
4.3.2 Model structural equation
4.3.3 Model statistics
4.3.4 Multicollinearity test
4.3.5 Comparative analysis
4.3.6 Projected influenza outbreak in US from 2021 to 2050
4.4 Estimation of the rate of deaths from influenza in USA
4.4.1 Fitted trend equation
4.4.2 Model statistics
4.4.3 Comparative analysis
4.4.4 Projected death rates for human casualty in USA due to influenza infection outbreak
4.5 Estimation of death resulting from flu infection in USA
4.5.1 Model name
4.5.2 Model structural equation
4.5.3 Model statistics
4.5.4 Multicollinearity test
4.5.5 Comparative analysis
4.5.6 Estimation of human casualty rate from influenza infection in USA
5 Discussion
6 Conclusion
Chapter 2. Toward equitable precision medicine: Investigating the transferability of deep learning models in clinical genetics across diverse populations
1 Introduction
1.1 Overview of precision medicine
1.2 Importance of equitable access and representation
1.3 Role of deep learning models in clinical genetics
1.4 Need for investigating model transferability across diverse populations
1.5 Purpose of the study
2 Overview of deep learning models in precision medicine
2.1 Existing research on model transferability
2.2 Disparities and challenges in clinical genetics research
2.3 Importance of addressing bias and representation
2.4 Current state of equitable precision medicine
3 Methodology
3.1 Search strategy and selection criteria
3.2 Data sources and study selection
3.3 Data extraction and synthesis
4 Nuances of model transferability across diverse populations
4.1 Overview of population diversity in clinical genetics
4.2 Factors influencing model performance across ethnicities
4.3 Case studies and examples of transferability challenges
4.4 Identification of bias and limitations in existing models
5 Strategies for improving model transferability
5.1 Data augmentation and SMOTE
5.2 Transfer learning and domain adaptation approaches
5.3 Addressing biases in data collection and model training
5.3.1 Data collection and sampling strategies
5.3.2 Bias mitigation during model training
5.3.3 Transfer learning and domain adaptation for reducing bias
5.3.4 Fairness metrics and evaluation
5.3.5 Collaborative data sharing and ethical considerations
5.4 Collaborative research and international data sharing initiatives
6 Application of transferable models in clinical practice
6.1 Case studies demonstrating successful model transferability
6.2 Impact on patient care and clinical decision-making
6.3 Challenges and considerations for implementation
6.4 Future directions for integrating transferable models into precision medicine
7 Ethical and societal implications
7.1 Ethical considerations in precision medicine research
7.2 Societal implications of equitable access to precision medicine
7.3 Addressing disparities in healthcare delivery
7.4 Policy recommendations for promoting equity in precision medicine
8 Conclusion
8.1 Summary of key findings
8.2 Contributions to equitable precision medicine
8.3 Recommendations for future research and practice
Chapter 3. Deep learning insights into transcriptomics and gene expression patterns analysis
1 Introduction
2 Deep learning in transcriptomics
2.1 Deep learning in transcriptomics using neural networks and models
2.2 Deep learning applications in spatial transcriptomics and single-cell RNA sequencing
3 Deep learning in gene expression analysis
3.1 The CMap project
3.2 Deep learning in disease classification
3.3 Challenges in predictive models with gene expression data
4 Conclusion
Chapter 4. Role of artificial intelligence in clinical cancer genomics for oncology
1 Introduction
2 History of sequencing platforms
3 Clinical genomics and its role in healthcare
4 Integration of AI in medicine using clinical genomics approach
5 Examples of AIM in oncology
6 Challenges and future potential
7 Conclusion
Chapter 5. Deep learning approaches for interpreting Non-coding regions in Ovarian cancer
1 Introduction
2 Background
2.1 Ovarian cancer
2.2 Ovarian cancer biomarkers
2.3 Noncoding regions in ovarian cancer
2.4 Genomic sequencing in interpreting noncoding regions
3 Deep learning in cancer
3.1 Deep learning for interpreting noncoding regions in ovarian cancer
3.2 Deep learning in healthcare technologies
4 Current development in ovarian cancer diagnosis
5 Current development in ovarian cancer prognosis
6 Conclusion
Chapter 6. Advancements in artificial intelligence-driven spatial transcriptomics: Decoding cellular complexity
1 Introduction
2 Fundamentals of spatial transcriptomics
2.1 Basics of transcriptomic
2.2 Spatial transcriptomics
2.3 Importance of spatial information in cellular analysis
3 Current challenges in spatial transcriptomics
3.1 Technical limitation
3.2 Data analysis bottlenecks
3.3 Interpretation challenges
4 Integration of AI in spatial transcriptomics
4.1 Machine learning and spatial transcriptomics
4.1.1 Convolutional neural networks
4.1.2 Graph neural networks
4.1.3 Variational autoencoders
4.1.4 Random forests
4.2 Deep learning approaches
4.3 Role of neural networks in data analysis
5 Advancements in data acquisition techniques
6 AI-driven data analysis pipelines
6.1 Preprocessing and quality control
6.2 Spatial clustering algorithms
6.3 Feature selection and dimensionality reduction
6.4 Visualization techniques
7 AI in decoding cellular complexity
7.1 Neural circuit mapping
7.2 Immune response profiling
7.3 Disease-specific case studies
8 Future directions and challenges
8.1 Ethical considerations
8.2 Unexplored frontiers in AI-driven spatial
9 Conclusion
Chapter 7. Advancements in clinical decision support through deep learning approaches in genetic diagnostics
1 Introduction
2 Role of advanced clinical decision support
2.1 Clinical decision support system
2.1.1 Machine learning in CDSS
3 Evolution of deep learning in healthcare
4 Clinical decision in genetic diagnostics
4.1 Genetic testing status
5 Deep learning architectures in genetic diagnosis
5.1 Convolutional neural networks
5.2 Recurrent neural networks
5.3 Other architectures
5.4 Hybrid deep learning model
6 Challenges in data heterogeneity with diverse genetic data sources
6.1 Interpretability importance in clinical decision-making
7 Precision medicine in clinical decision-making
7.1 Expanding applications in precision medicine
8 Discussion
9 Conclusion
Chapter 8. Neural architectures for genomic understanding: Deep dive into epigenome and chromatin structure
1 Introduction
2 Neural architectures for genomic understanding
2.1 Different types of data handled by neural architectures
3 Application of CNNs in genomic sequences
3.1 CNNs in capturing spatial dependencies in genomic sequences
3.2 CNN applications in genomic data analysis
3.3 Spatial organization of the epigenome
4 RNNs for temporal dynamics in epigenome
4.1 RNNs for modeling sequential dependencies
4.2 RNNs in capturing temporal dynamics of chromatin structure
4.3 RNNs in analyzing time-series data related to gene expression
5 GNNs for representing chromatin structure as a graph
6 Integrating CNNs, RNNs, and GNNs for comprehensive genomic analysis
7 Conclusion
List of abbreviations
Chapter 9. Deep learning in personalized genomics and gene editing
1 Introduction
2 Introduction to artificial intelligence
3 Personalized medicine
4 Drug response prediction
5 Cancer molecular subtyping
6 Classification of cancer of unknown primary
7 Survival/prognosis prediction
8 Synthetic lethality prediction
9 Genome editing
10 Translating preclinical animal models to human trials
11 Conclusion
Chapter 10. Deep learning–based model for prediction of prognostic genes of breast cancer using transcriptomic data
1 Introduction
1.1 Risk factors
1.2 Diagnosis and current treatment strategies
1.3 Transcriptomic data analysis in cancer research
1.4 Artificial Intelligence and Deep Learning
1.5 Application of deep learning in transcriptomic data analysis
2 Materials and methodology
2.1 TCGA BRCA mRNA-Seq dataset selection
2.2 Data validation and preprocessing
2.3 Identification of the differentially expressed genes
2.3.1 Deep learning–based most significant gene subset identification
2.3.2 Preparation of dataset for the deep learning classifier
2.3.3 Classification model
2.3.4 Performance evaluation of the DNN classifier
2.3.5 Protein–protein network analysis and identification of significant module
3 Results
3.1 Differentially expressed gene identification
3.2 Most significant gene subsets
3.3 Overlapping genes between DEGs and the DNN-classified gene subsets
3.4 Gene ontology of common genes
3.5 PPI network
3.6 Significant module and its associated genes
4 Discussion
5 Conclusions
6 Limitations
Chapter 11. Genomic image analysis: Bridging genomics and advanced imaging
1 Introduction
1.1 Related work
2 Fundamentals of genomic imaging
2.1 Techniques in genomic imaging
2.2 Applications of genomic imaging
2.3 Challenges in genomic imaging
3 Image processing techniques
3.1 Preprocessing in GIA
3.1.1 Noise reduction
3.1.2 Normalization
3.1.3 Image enhancement
3.1.4 Artifact removal
3.1.5 Resolution enhancement
3.2 Feature extraction in genomic images
3.2.1 Segmentation
3.2.2 Pattern recognition
3.3 Data representation in GIA
3.3.1 Pixel intensity matrices
3.3.2 Feature vectors
3.3.3 Hybrid approaches
4 Data analysis and interpretation
4.1 Statistical analysis
4.2 Statistical methods in GIA
4.2.1 Hypothesis testing
4.3 Computational models
4.4 Visualization techniques
5 Innovations in genomic image analysis
5.1 Disease diagnosis
5.2 Drug discovery
5.3 Personalized medicine
6 Challenges and future directions
6.1 Data complexity and volume
6.2 Ethical and privacy concerns
6.3 Integration with clinical practice
6.4 Future trends
7 Conclusion
Chapter 12. Qualitative study on steganography of genomic image data for secure data transmission using deep learning models
1 Introduction
2 Deep learning framework and its applications in genomic imaging
3 Creation of genomic images from DNA
4 Works related to DNA-based steganography
5 Steganography of genomic data using classical and deep learning techniques
5.1 Least significant bit method
5.2 Discrete cosine transform
5.3 Spread spectrum techniques
5.4 Deep learning using autoencoders
5.5 Deep learning using convolutional neural networks
5.6 Deep learning using generative adversarial networks
6 Conclusion and future works
Chapter 13. Generative artificial intelligence in genetics: A comprehensive review
1 Introduction
2 Methods
2.1 The early advent of artificial intelligence in genetics
2.2 Bayesian inference
2.3 Recent advancements in generative artificial intelligence in genetics
2.4 Explainable artificial intelligence methods for generative AI models
2.5 AlphaFold and its implications for generative AI in genetics
2.6 Nonprotein applications of generative artificial intelligence
3 Conclusion
Chapter 14. Integrating computational biology and multiomics data for precision medicine in personalized cancer treatment
1 Introduction
2 Personalized medicine in cancer treatment
2.1 Role of cancer genomics in precision medicine
2.2 Genomic changes in metastasis
2.3 Examples of precision medicine of cancer genomics
2.4 Molecular classification of various cancers and its implications for personalized therapeutic approaches
3 Next-generation sequencing in cancer genomics
3.1 Overview of next-generation sequencing
3.2 NGS revolutionizing precision cancer treatment across diverse tumor types
3.3 Challenges and future perspectives
4 Role of computational analysis in cancer genomics
4.1 Computational methods or tools for cancer genome interpretation
4.2 Challenges and future perspectives
5 Integration of computational biology and multi-omics data
5.1 Multiomics approaches in cancer research
5.2 Computational frameworks for multi-omics studies
5.3 Multiomics data integration methods
5.3.1 Network-based methods
5.4 Advancement of multiomics data in cancer therapeutics
6 Conclusion
Chapter 15. Deep generative models in utilitarian and metamorphic genomics—Intellectual benefits
1 Introduction
2 Deep utilitarian models
2.1 Introduction to various generative models
2.1.1 An overview of GANs
2.1.2 An overview of variational autoencoders
2.2 Significance of GANs in the deep learning
2.2.1 Training in adversarial settings
2.2.2 Application in generating synthetic data
2.3 Probabilistic nature of variational autoencoders
2.3.1 Neural network components: Encoder and decoder
2.3.2 Continuous latent spaces and their implications for generative modeling
2.4 Network architectures in deep generative models
2.4.1 Fully connected, recurrent, and convolutional neural networks
2.4.2 Adaptation based on data nature, task, and computational resources
3 The generation of genomic data
3.1 Applications in utilitarian genomics
3.1.1 Design of Useful Sequences with Preferred Characteristics
3.1.2 Integration of biological domain knowledge
3.2 Application in metamorphic biology and population genetics
3.2.1 Role of biobanks in evolutionary research and disease associations
3.2.2 Challenges in accessing genomic data for research
4 Dimensionality reduction and visualization
4.1 Applications in utilitarian genomics
4.1.1 DGM-based dimensionality reduction in transcriptomic data
4.1.2 Cell-type clustering and classification using latent space
4.2 Application in metamorphic biology and the population genetics
4.2.1 Capturing fine population structure in SNP data
4.2.2 Training models on real and samples that were simulated and had known demographic history
5 Conclusions
Chapter 16. Bridging the gap: Understanding genetic discoveries through explainable artificial intelligence
1 Introduction
1.1 Explainable artificial intelligence
1.1.1 The role of XAI in genetic data evaluation
1.1.2 XAI in action
2 Interpreting genetic data with XAI
2.1 The foundation of XAI stands on three pillars
2.2 Types of genetic data
2.2.1 Genome-wide association studies
2.2.2 Whole-genome sequencing
2.2.3 Expression data
2.2.4 Epigenetic data
2.2.5 Microbiome data
2.3 Overcoming difficulties with interpretability
3 Biases and ethical considerations
4 Strategies for XAI models
5 The future of human-AI collaboration
6 Conclusion
Chapter 17. Explainable Artificial Intelligence in genetics: A case study
1 Introduction
2 The need for explainable AI in genetics
3 Foundations of explainable AI
4 XAI techniques for genetic analysis
4.1 Model-specific approaches
4.2 Model-agnostic approaches
4.3 Probing technique
4.4 Perturbing technique
4.5 Surrogate Technique
4.6 Global explanations
4.7 Local explanations
4.8 XAI tools/libraries
5 Case studies in genetic interpretability
5.1 Case study 1
5.2 Case study 2
5.3 Limitations of the XAI methods and comparison with other XAI techniques
6 Advantages and challenges of integrating XAI in genetics
6.1 Advantages of integrating XAI in genetics
6.2 Challenges of integrating XAI in genetics
6.3 Addressing potential challenges and limitations of XAI
7 Future directions and implications
7.1 The potential influence of XAI on genetic research
7.2 Ethical, legal, and societal implications of XAI in genetics
7.3 Ethical implications
7.4 Legal implications
7.5 Societal implications and future directions
8 Conclusion
Chapter 18. Deep learning in predicting genetic disorders: A case study of diabetic kidney disease
1 Introduction
1.1 Biological data and deployment of AI systems
2 Background of deep learning
2.1 Artificial neural networks
2.2 Convolution neural network
2.3 Recurrent neural network
2.4 Long short-term memory
2.5 Autoencoder
2.6 Restricted Boltzmann machine
2.7 Gated recurrent unit
3 Relationships between genetics and diabetic kidney disease
4 Application of deep learning models in detecting diabetic kidney disease
4.1 Study findings
4.2 Imitations of the deep learning models for DKD prediction
4.3 Challenges and consideration
4.4 Future prospective in predictive DKD
5 Conclusion
Chapter 19. Artificial intelligence and deep learning in single-cell omics data analysis: A case study
1 Introduction
1.1 Overview of single-cell technologies
1.2 Importance of single-cell omics in genetics and genomics research
1.3 Possible applications of single-cell omics
1.4 Challenges in analyzing single-cell data
2 Fundamentals of deep learning
2.1 Basics of artificial neural networks
2.2 Deep learning architectures
2.3 Training and optimization algorithms
3 Single-cell data analysis pipeline
3.1 Preprocessing single-cell omics data
3.1.1 Removal of low-quality data
3.1.2 Normalization
3.1.3 Batch correction
3.1.4 Feature selection
3.1.5 Dimensionality reduction
3.1.6 Clustering
3.1.7 Differential expression analysis
3.2 Quality control and filtering
3.3 Normalization techniques
3.4 Dimensionality reduction methods
3.5 Down-stream analysis with deep learning
4 Deep learning applications in single-cell omics analysis
4.1 Cell type identification and classification
4.2 Clustering and cell trajectory inference
4.3 Differential expression analysis
4.4 GRN reconstruction
4.5 Spatial transcriptomics analysis
5 Advanced deep learning models for single-cell omics
5.1 Graph neural networks for single-cell data
5.2 Attention mechanism in single-cell data analysis
5.3 Deep generative models for single-cell data synthesis
5.4 Integration of multiomics data with deep learning
6 Case studies and applications
6.1 Multimodal single-cell data analysis
6.2 Transfer learning approaches for multiomics integration
6.3 Cell-type prediction using deep learning in single-cell omics
6.4 Single-cell gene regulatory network inference
7 Challenges and future directions
7.1 Interpretability and explainability of deep learning models
7.2 Scalability and computational challenges
7.3 Advanced multiomics integration
7.4 Enhanced interpretability of AI models
7.5 Integrating spatial and temporal dynamics
7.6 Biologically informed deep learning architecture
7.7 Real-time single-cell analysis
7.8 Precision medicine and disease detection
8 Conclusion
List of abbreviations
Declaration of generative AI contents
Chapter 20. Deep learning for network building and analysis of biological networks: A case study
1 Introduction
2 Biological networks
2.1 Biological network in disease prediction and treatment
3 Deep learning and its application in biological networks
4 Limitations and challenges
5 Case study
5.1 Proteomics
5.2 Drug development, discovery, and polypharmacy
5.3 Computational pathology
5.4 Disease diagnosis
6 Conclusion
Chapter 21. Transformer networks and autoencoders in genomics and genetic data interpretation: A case study
1 Introduction
1.1 Background and context
1.2 Purpose of the chapter
1.3 Significance of integrating transformer networks and autoencoders in genomics
2 The evolving landscape of genetic data interpretation
2.1 Overview of genetic data interpretation challenges
2.2 Traditional methods and challenges in genetic data interpretation
3 Transformer networks in genomic data analysis
3.1 Role in sequential data analysis
3.2 DNABERT pipeline
3.3 Significance in capturing long-range dependencies in genetic sequences
3.4 Comparative analysis with traditional methods
4 Autoencoders in genomic data analysis: Unraveling the genetic code through dimensionality reduction, feature learning, and efficient compression
4.1 Applications in dimensionality reduction and feature learning
4.2 Efficient genomic data compression using autoencoders
4.3 Evaluating the fidelity of reconstructed genetic data
5 Genomic data representation challenges
5.1 Complexity of genomic data representation
5.2 Importance of effective representation for analysis
5.3 Role of transformer networks in addressing representation challenges
6 Case studies in feature extraction
6.1 Application of transformer networks for feature extraction
6.1.1 Introduction to Transformer Networks in genomics
6.1.2 Sequential data analysis in genomics
6.1.3 Application scenarios
6.1.4 Case study 1: Identification of disease-associated mutations
6.1.5 Case study 2: Enhancer–promoter interaction prediction
6.2 Comparative analysis
6.3 Comparative analysis with traditional methods
6.3.1 Traditional methods in feature extraction
6.3.2 Performance metrics
6.3.3 Case studies illustrating differences
6.3.4 Case study 3: Comparative analysis of gene expression prediction
6.4 Utilizing autoencoders for compressing and reconstructing genetic data
6.4.1 Introduction to autoencoders in genomics
6.4.2 Dimensionality reduction and feature learning
6.4.3 Case studies in compression and reconstruction
6.4.4 Case study 4: Compression of whole-genome sequences
6.4.5 Case study 5: Reconstruction of genetic variants
6.5 Evaluation of the fidelity of reconstructed data
6.5.1 Metrics for data fidelity
6.5.2 Comparison with ground truth
6.5.3 Implications for genomic analysis
7 Integration of Transformer Networks and Autoencoders in genomics: Advantages, case studies, and implications for precision medicine and personalized genomics
7.1 Benefits of integrating both technologies
7.1.1 Comprehensive feature extraction
7.1.2 Enhanced data representation
7.1.3 Improved reconstruction fidelity
7.2 Impact on precision medicine and personalized genomics
7.2.1 Precision medicine advancements
7.2.2 Personalized genomics tailoring
8 Future trends and possibilities
8.1 Emerging trends in genomics and data interpretation
8.1.1 Multimodal data integration
8.1.2 Explainable AI in genomics
8.2 Potential advancements in Transformer Networks and Autoencoders
8.2.1 Attention mechanism refinements
8.2.2 Unsupervised learning strategies for autoencoders
8.3 Implications for the future of genomics research
8.3.1 Precision medicine advancements
8.3.2 Accelerated drug discovery
8.3.3 Informatics in clinical genomics
9 Challenges and potential limitations
9.1 Identification of challenges in integrating Transformer Networks and Autoencoders
9.2 Possible limitations of the combined approach
9.3 Strategies to address challenges and mitigate limitations
10 Conclusion
11 Declaration of generative AI and AI-assisted technologies in the writing process
Index

Life Sciences

Physical Sciences & Engineering

Social Sciences & Humanities

Health

Deep Learning in Genetics and Genomics

Volume 2: Advanced Applications

Purchase options

World Book Day Celebration!

Institutional subscription on ScienceDirect

Khalid Raza

Related books