Biological Data Integration

Christine Froidevaux is Professor Emeritus of Computer Science at the Université Paris-Saclay, France, and her research focuses on data integration for biological systems.Marie-Laure Martin-Magniette is Research Director at INRAE, France, where she develops statistical methodologies for analyzing and integrating omics data.Guillem Rigaill is Research Director at INRAE, France, where he develops statistical methodologies for (multi)omics analysis.

• Preface xiChristine FROIDEVAUX, Marie-Laure MARTIN-MAGNIETTE and Guillem RIGAILLPart 1 Knowledge Integration 1Chapter 1 Clinical Data Warehouses 3Maxime WACK and Bastien RANCE1.1 Introduction to clinical information systems and biomedical warehousing: data warehouses for what purposes? 31.1.1 Warehouse history 41.1.2 Using data warehouses today 41.2 Challenge: widely scattered data 51.3 Data warehouses and clinical data 61.3.1 Warehouse structures 61.3.2 Warehouse construction and supply 111.3.3 Uses 111.4 Warehouses and omics data: challenges 151.4.1 Challenges of data volumetry and structuring omic data 161.4.2 Attempted solutions 171.5 Challenges and prospects 181.5.1 Toward general-purpose warehouses 181.5.2 Ethical dimension of the implementation and the use of warehouses 191.5.3 Origin and reproducibility 191.5.4 Data quality 201.5.5 Data warehousing federation and data sharing 211.6 References 21Chapter 2 Semantic Web Methods for Data Integration in Life Sciences 25Olivier DAMERON2.1 Data-related requirements in life sciences 262.1.1 Databases for the life sciences 262.1.2 Requirements 272.1.3 Common approaches: InterMine and BioMart 302.2 Semantic Web 312.2.1 Techniques 322.2.2 Implementation 422.3 Perspectives 432.3.1 Facilitating appropriation to users 432.3.2 Facilitating the appropriation by software programs: FAIR data 442.3.3 Federated queries 452.4 Conclusion 462.5 References 47Chapter 3 Workflows for Bioinformatics Data Integration 53Sarah COHEN-BOULAKIA and Frédéric LEMOINE3.1 Introduction 533.2 Bioinformatics data processing chains: difficulties 543.2.1 Designing a data processing chain 553.2.2 Analysis execution and reproducibility 563.2.3 Maintenance, sharing and reuse 583.3 Solutions provided by scientific workflow systems 593.3.1 Fundamentals of workflow systems 593.3.2 Workflow systems 643.4 Use case: RNA-seq data analysis 693.4.1 Study description 693.4.2 From data processing chain to workflows 723.4.3 Data processing chains implemented as workflows: conclusion 753.5 Challenges, open problems and research opportunities 773.5.1 Formalizing workflow development 773.5.2 Workflow testing 783.5.3 Discovering and sharing workflows 793.6 Conclusion 803.7 References 81Part 2 Integration and Statistics 87Chapter 4 Variable Selection in the General Linear Model: Application to Multiomic Approaches for the Study of Seed Quality 89Céline LÉVY-LEDUC, Marie PERROT-DOCKÈS, Gwendal CUEFF and Loïc RAJJOU4.1 Introduction 904.2 Methodology 934.2.1 Estimation of the covariance matrix Σq 934.2.2 Estimation of B 964.3 Numerical experiments 994.3.1 Statistical performance 994.3.2 Numerical performance 1004.4 Application to the study of seed quality 1034.4.1 Metabolomics data 1044.4.2 Proteomics data 1054.5 Conclusion 1084.6 Appendices 1084.6.1 Example of using the package MultiVarSel for metabolomic data analysis 1084.6.2 Example of using the package MultiVarSel for proteomic data analysis 1104.7 Acknowledgments 1134.8 References 113Chapter 5 Structured Compression of Genetic Information and Genome-Wide Association Study by Additive Models 117Florent GUINOT, Marie SZAFRANSKI and Christophe AMBROISE5.1 Genome-wide association studies 1185.1.1 Introduction to genetic mapping and linkage analysis 1185.1.2 Principles of genome-wide association studies 1195.1.3 Single nucleotide polymorphism 1205.1.4 Disease penetrance and odds ratio 1225.1.5 Single marker analysis 1245.1.6 Multi-marker analysis 1265.2 Structured compression and association study 1325.2.1 Context 1325.2.2 New structured compression approach 1335.3 Application to ankylosing spondylitis (AS) 1425.3.1 Data 1425.3.2 Predictive power evaluation 1435.3.3 Manhattan diagram 1445.3.4 Estimation for the most significant SNP aggregates 1445.4 Conclusion 1465.5 References 146Chapter 6 Kernels for Omics 151Jérôme MARIETTE and Nathalie VIALANEIX6.1 Introduction 1526.2 Relational data 1536.2.1 Data described by the kernel 1536.2.2 Data described by a general (dis)similarity measure 1556.3 Exploratory analysis for relational data 1586.3.1 Kernel clustering 1586.3.2 Kernel principal component analysis 1616.3.3 Kernel self-organizing maps 1636.3.4 Limitations of relational methods 1666.4 Combining relational data 1686.4.1 Data integration in systems biology 1686.4.2 Kernel approaches in data integration 1696.4.3 A consensual kernel 1726.4.4 A parsimonious kernel that preserves the topology of the initial data 1736.4.5 A complete kernel preserving the topology of the initial data 1756.5 Application 1766.5.1 Loading Tara Ocean data 1766.5.2 Data integration by kernel approaches 1776.5.3 Exploratory analysis: kernel PCA 1796.6 Session information for the results of the example 1866.7 References 188Chapter 7 Multivariate Models for Data Integration and Biomarker Selection in ‘Omics Data 195Sébastien DÉJEAN and Kim-Anh LÊ CAO7.1 Introduction 1957.2 Background 1977.2.1 Mathematical notations 1977.2.2 Terminology 1987.2.3 Multivariate projection-based approaches 1987.2.4 A criterion to maximize specific to each methodology 1997.2.5 A linear combination of variables to reduce the dimension of the data 1997.2.6 Identifying a subset of relevant molecular features 2007.2.7 Summary 2007.3 From the biological question to the statistical analysis 2017.3.1 Exploration of one dataset: PCA 2017.3.2 Classify samples: projection to latent structure discriminant analysis 2067.3.3 Integration of two datasets: projection to latent structure and related methods 2107.3.4 Integration of several datasets: multi-block approaches 2157.4 Graphical outputs 2207.4.1 Individual plots 2207.4.2 Variable plots 2217.5 Overall summary 2227.6 Liver toxicity study 2237.6.1 The datasets 2237.6.2 Biological questions and statistical methods 2237.6.3 Single dataset analysis 2247.6.4 Integrative analysis 2317.7 Conclusion 2387.8 Acknowledgments 2387.9 Appendix: reproducible R code 2397.9.1 Toy examples 2397.9.2 Liver toxicity 2437.10 References 247List of Authors 251Index 255