Discover millions of ebooks, audiobooks, and so much more with a free trial

Only $11.99/month after trial. Cancel anytime.

Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research
Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research
Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research
Ebook1,483 pages14 hours

Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research

Rating: 0 out of 5 stars

()

Read preview

About this ebook

Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research presents the most exciting molecular and recombinant DNA techniques used in the analysis of brain function and behavior, a critical piece of the puzzle for clinicians, scientists, course instructors and advanced undergraduate and graduate students. Chapters examine neuroinformatics, genetic and neurobehavioral databases and data mining, also providing an analysis of natural genetic variation and principles and applications of forward (mutagenesis) and reverse genetics (gene targeting). In addition, the book discusses gene expression and its role in brain function and behavior, along with ethical issues in the use of animals in genetics testing.

Written and edited by leading international experts, this book provides a clear presentation of the frontiers of basic research as well as translationally relevant techniques that are used by neurobehavioral geneticists.

  • Focuses on new techniques, including electrocorticography, functional mapping, stereo EEG, motor evoked potentials, optical coherence tomography, magnetoencephalography, laser evoked potentials, transmagnetic stimulation, and motor evoked potentials
  • Presents the most exciting molecular and recombinant DNA techniques used in the analysis of brain function and behavior
  • Written and edited by leading international experts
LanguageEnglish
Release dateApr 24, 2018
ISBN9780128041161
Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research

Related to Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research

Related ebooks

Biology For You

View More

Related articles

Related categories

Reviews for Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research

Rating: 0 out of 5 stars
0 ratings

0 ratings0 reviews

What did you think?

Tap to rate

Review must be at least 10 words

    Book preview

    Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research - Robert T. Gerlai

    Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research

    Editor

    Robert T. Gerlai

    Table of Contents

    Cover image

    Title page

    Copyright

    Dedication

    Contributors

    Preface

    Acknowledgments

    Why Should We Use Genetics in the Analysis of Brain Function and Behavior? Practical and Theoretical Considerations

    Section I. Neuroinformatics, Computational Models and Data Analysis

    Chapter 1. Neurological Biomarkers and Neuroinformatics: The Role of The Virtual Brain

    Adding Information to Brain Imaging Data Via Brain Network Modeling

    Bridging Scales Within The Virtual Brain

    Pragmatics of Modeling in The Virtual Brain

    Modeling With The Virtual Brain in the Clinical Context

    Summary: The Role of Neuroinformatics in Biomarker Discovery

    Glossary of Terms

    Chapter 2. Informatics for Interoperability of Molecular-Genetic and Neurobehavioral Databases

    Introduction

    Methods

    Discussion

    Chapter 3. The Allen Brain Atlas: Toward Understanding Brain Behavior and Function Through Data Acquisition, Visualization, Analysis, and Integration

    Mapping Gene Expression in the Brain

    Genetic Tools: Transgenic Mouse Lines

    Community Use of the Allen Institute for Brain Science Resources

    Conclusion

    Chapter 4. ExAtlas: Online Tool to Integrate Gene Expression and Gene Set Enrichment Analyses

    Introduction

    Experimental Approaches to Gene Expression Profiling

    Preprocessing of Gene Expression Data

    Overview of the Statistical Analysis of Gene Expression Profiles in ExAtlas

    Data Management in ExAtlas

    Analysis of a Single Data Set With Gene Expression Profiles

    Correlation Between Gene Expression Profiles

    Gene Set Overlap and Gene Set Enrichment

    Meta-Analysis of Paired Gene Expression Profiles

    Exercises (Tutorial)

    Conclusions

    Chapter 5. Computational Models: How Do They Help to Understand Neurologic Diseases?

    Introduction

    The Challenge of Modeling Brain Complexity

    Computational Models and Their Correspondence to Complex Biologic Reality

    What Are Our Readouts of Neurologic Disorders?

    Neurologic Disorders From Local to Large-Scale Networks

    Which Computational Models Are Useful in Neurology and Why?

    Modeling Alzheimer Disease

    Models of Parkinson Disease

    Models of Down Syndrome

    Conclusions

    Chapter 6. The Use of Recombinant Inbred Strains in Systems Genetics and Functional Analyses in Behavioral Pharmacology

    Introduction

    Dopamine D2 Receptor Densities in the Dorsal Striatum

    Conclusions

    Section II. Searching for New Genes: Natural Genetic Variation

    Chapter 7. Spontaneous Versus Induced Mutations: Conceptual and Methodological Considerations for the Neurobehavioral Geneticist

    Introduction

    Utilizing Naturally Occurring Variance: The Old School of Behavior Genetics?

    Analysis of Induced Mutations: 21st-Century Recombinant DNA Technology and Its Advantages

    Summary and Conclusions

    Chapter 8. Reduced Complexity Cross Design for Behavioral Genetics

    Introduction

    Trait Selection

    Genotyping Strategies

    QTL Mapping and QTG Discovery in the RCC

    Conclusion

    Chapter 9. Collaborative Cross as the Next-Generation Mouse Genetic Reference Population Designed for Dissecting Complex Traits

    Background

    Communicable Diseases

    Non-Communicable Diseases Mapping Host Susceptibility to Intestinal Cancer

    Collaborative Cross Lines for Type 2 Diabetes Study

    Collaborative Cross Lines for Fundamental Host Body Traits

    Collaborative Cross Lines for Next-Generation Pharmacogenetics

    Conclusions

    Chapter 10. Discovery of Novel Genes and Other Lineage-Specific Features Through Comparative Genomics

    Introduction

    Strategies for Identifying Lineage-Specific Expansions and De Novo Genes

    Identifying De Novo Genes

    High-Throughput Strategies for Identifying Novel Genes

    Predicting the Function(s) of Novel Genes

    Conclusions

    Chapter 11. Using Signatures of Directional Selection to Guide Discovery

    Introduction

    Selective Breeding: The Basics

    Genetic and Genomic Consequences of Selection: QTL and Expression Arrays

    Genetic and Genomic Consequences of Selection: Beyond Gene Lists

    Genetic and Genomic Consequences of Selection: Beyond Rats and Mice

    Utility of Selection Signatures in the Future

    Section III. Discovery of New Genes and New Functions of Genes Using Gene Expression Analyses

    Chapter 12. Using Transcriptomics to Study Behavior

    Introduction

    Methodology

    Summary

    Chapter 13. Genes, Behavior, and Next-Generation Sequencing: The First 10 Years

    Introduction

    The Early Years: Microarrays to RNA-Seq

    The Early Years: Development of RNA-Seq

    Recent Progress in RNA-Seq, the Brain, and Behavior

    Conclusions

    Chapter 14. Abnormal Social Behaviors and Dysfunction of Autism-Related Genes Associated With Daily Agonistic Interactions in Mice

    Introduction

    Materials and Methods

    Behavioral Analysis

    Results

    Discussion

    Conclusion

    Abbreviations

    Funding

    Author Contributions

    Chapter 15. Epigenetic Mechanisms of Learning and Memory

    Introduction: Memory, Gene Regulation, and Epigenetics

    Epigenetic Mechanisms

    Epigenetic Mechanisms Regulating Memory

    Epitranscriptomics: The New Horizon of RNA Epigenetics

    Integrating the Epigenetic Code to Form and Recall Memories

    High-Throughput Next-Generation Sequencing: From Candidate Genes to Genome-Wide Analysis

    Conclusions

    Section IV. Discovery of Genes and Biological Mechanisms: Forward Genetics and Other Screening-Based Methods

    Chapter 16. Systematic Screens in Zebrafish Shed Light on Cellular and Molecular Mechanisms of Complex Brain Phenotypes


    Introduction

    Strategies for Forward, Reverse, and Chemical Genetic Screens

    Generation of a Library of Induced Genetic or Epigenetic Variations

    Postscreening: Efforts to Discover Underlying Cellular and Molecular Mechanisms

    Examples of Screens in Zebrafish

    Uncovering Cellular and Molecular Mechanisms Through Studying Natural Phenotypic Variations

    Conclusion

    Chapter 17. The Transition of Zebrafish Functional Genetics From Random Mutagenesis to Targeted Integration

    Introduction

    Zebrafish Random Mutagenesis

    Targeted Modifications

    Applications of Gene Editing

    Conclusion

    Section V. Manipulating Known Genes to Understand Biological Function: Reverse Genetics

    Chapter 18. Molecular Techniques Used to Explore Glutamate Receptors in Synaptic Plasticity and Memory

    Introduction

    Part 1: RNA Interference

    Part 2: Knockout and Conditional Knockout

    Part 3: Knock-In

    Part 4: CRISPR

    Conclusion

    Chapter 19. Using Herpes Simplex Virus Type 1-Based Amplicon Vectors for Neuroscience Research and Gene Therapy of Neurologic Diseases

    Introduction

    Herpes Simplex Virus Type 1

    Brief Introduction to Herpes Simplex Virus Type 1-Based Vectors

    Herpes Simplex Virus Type 1-Based Amplicon Vectors

    Amplicon Vectors in Neuroscience (Fig. 19.6)

    Amplicon Vectors in Experimental Analysis of Behavior, Learning, and Memory

    Neurodegenerative Diseases

    Concluding Remarks

    Chapter 20. Cre-lox Neurogenetics: History, Present, and Future

    Introduction

    Young and Foolish

    Cre-Driver Resources for Neurosciences

    Cre-Driving Into the Future

    Chapter 21. Functional Analysis of Proteins Involved in Neurodegeneration Using the Model Organism Dictyostelium: Alzheimer's, Huntington's, and Batten Disease

    Introduction

    Alzheimer's Disease

    Huntington's Disease

    Neuronal Ceroid Lipofuscinosis (Batten Disease)

    The Value and Limitations of Dictyostelium as a Biomedical Model

    Conclusion

    Abbreviations

    Chapter 22. The Role of Human Endogenous Retroviruses (HERVs) in the Pathologies of the Nervous System

    Introduction

    Human Endogenous Retroviruses in Neurological Diseases

    Human Endogenous Retroviruses in Psychiatric Diseases

    Conclusions

    Chapter 23. Optogenetics Dissection of Sleep Circuits and Functions

    Introduction

    Principles for Optogenetic Dissection of Neural Activity

    Optogenetic Dissection of Sleep Circuits

    Perspectives

    Chapter 24. The Use of DREADDs for Dissecting the Contribution of Cellular and Neural Circuit Mechanisms in Models of Neurodegenerative Disease

    Designer Receptors Exclusively Activated by Designer Drugs

    Neurodegenerative Disease

    Translational Potential of Chemogenetics for Treating Neurodegenerative Disease

    Abbreviations

    Section VI. Ethical Considerations

    Chapter 25. Genes and Human Behavior: Ethical Implications

    Introduction

    Henry Herbert Goddard and the Kallikak Family

    The Rise and Fall of the Criminal Chromosome

    Monoamine Oxidase A: The Warrior Gene?

    Boys, Girls, and Mathematical Ability

    Smart Genes, Stupid Genes

    Biologic Race Revisited or Not

    Conclusion

    Chapter 26. Ethical Considerations for Animal Use in Behavioral and Neural Research

    Introduction

    Ethical Considerations for Animal Use in Science

    The 3Rs

    Humane Endpoints

    Regulatory Oversight Framework

    IACUCs

    Ethics Review of Proposed Animal Use

    Planning Animal Use Protocols

    Degree of Invasiveness

    Determining Animal Numbers Required

    Descriptions of Experimental and Animal Care Details

    Standard Operating Procedures

    Adverse Effects and Monitoring

    Animal Models: Welfare, Ethics, and Safety

    Breeding Animals

    Euthanasia

    Postapproval Monitoring

    Conclusion

    Index

    Copyright

    Academic Press is an imprint of Elsevier

    125 London Wall, London EC2Y 5AS, United Kingdom

    525 B Street, Suite 1800, San Diego, CA 92101-4495, United States

    50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

    The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

    Copyright © 2018 Elsevier Inc. All rights reserved.

    No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

    This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

    Notices

    Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

    Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

    To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

    Library of Congress Cataloging-in-Publication Data

    A catalog record for this book is available from the Library of Congress

    British Library Cataloguing-in-Publication Data

    A catalogue record for this book is available from the British Library

    ISBN: 978-0-12-804078-2

    For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

    Publisher: Nikki Levy

    Acquisition Editor: Natalie Farra

    Editorial Project Manager: Kristi Anderson

    Production Project Manager: Poulouse Joseph

    Designer: Robert T. Gerlai

    Typeset by TNQ Books and Journals

    Dedication

    In memory of Dr. John C. Roder (July 17, 1950–January 6, 2018), a pioneer of modern neurobehavioral genetics, my mentor and friend.

    Contributors

    Hanifa J. Abu Toamih Atamni,     Tel Aviv University, Tel Aviv, Israel

    Antoine Adamantidis,     Department for BioMedical Research, University of Bern, Bern, Switzerland

    Eva Adamová,     The University of Tennessee Health Science Center, Memphis, TN, United States

    Alejandra I. Aguirre,     IBCN, University of Buenos Aires and CONICET, Ciudad de Buenos Aires, Argentina

    Vladimir N. Babenko,     Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Novosibirsk, Russia

    Maria V. Baez,     IBCN, University of Buenos Aires and CONICET, Ciudad de Buenos Aires, Argentina

    Jon Beckwith,     Harvard Medical School, Boston, MA, United States

    Douglas M. Bowden,     National Primate Research Center at the University of Washington, Seattle, WA, United States

    Camron D. Bryant,     Boston, MA, United States

    Jarryd M. Campbell,     Mayo Clinic, Rochester, MN, United States

    Karl J. Clark,     Mayo Clinic, Rochester, MN, United States

    John C. Crabbe,     Oregon Health & Science University, and VA Portland Health Care System, Portland, OR, United States

    M. Imad Damaj,     Richmond, VA, United States

    Priscila Darakjian,     Oregon Health & Science University, Portland, OR, United States

    Fernando P.M. De Villena,     University of North Carolina at Chapel Hill, Chapel Hill, NC, United States

    Mara Dierssen

    Centre for Genomic Regulation (CRG), Barcelona, Spain

    Universitat Pompeu Fabra (UPF), Barcelona, Spain

    CIBER de Enfermedades Raras (CIBERER), Barcelona, Spain

    Evan Dong,     National Primate Research Center at the University of Washington, Seattle, WA, United States

    Mark F. Dubach,     National Primate Research Center at the University of Washington, Seattle, WA, United States

    Louis Y. El Khoury,     Mayo Clinic, Rochester, MN, United States

    Alberto L. Epstein,     UMR INSERM U1179 – Université de Versailles Saint Quentin en Yvelines, UFR des Sciences de la Santé Simone Veil, Montigny le Bretonneux, France

    Martin T. Ferris,     University of North Carolina at Chapel Hill, Chapel Hill, NC, United States

    Anna G. Galyamina,     Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Novosibirsk, Russia

    Robert T. Gerlai,     University of Toronto Mississauga, Mississauga, ON, Canada

    Terri L. Gilbert,     GilHou Science Communication, Seattle, WA, United States

    Su Guo,     University of California San Francisco, San Francisco, CA, United States

    David Hanwell,     University of Toronto, Toronto, ON, Canada

    Christina A. Harrington,     Oregon Health & Science University, Portland, OR, United States

    Robert Hitzemann

    Oregon Health & Science University, Portland, OR, United States

    Veterans Affairs Medical Center, Portland, OR, United States

    Robert J. Huber,     Trent University, Peterborough, ON, Canada

    Ovidiu D. Iancu,     Oregon Health & Science University, Portland, OR, United States

    Fuad A. Iraqi,     Tel Aviv University, Tel Aviv, Israel

    Diana A. Jerusalinsky,     IBCN, University of Buenos Aires and CONICET, Ciudad de Buenos Aires, Argentina

    Zhengping Jia

    The Hospital for Sick Children, Toronto, ON, Canada

    University of Toronto, Toronto, ON, Canada

    Byron C. Jones,     The University of Tennessee Health Science Center, Memphis, TN, United States

    Sulev Kõks

    University of Tartu, Tartu, Estonia

    Estonian University of Life Sciences, Tartu, Estonia

    Prion OÜ, Tartu, Estonia

    Gea Kõks

    University of Tartu, Tartu, Estonia

    Prion OÜ, Tartu, Estonia

    Irina L. Kovalenko,     Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Novosibirsk, Russia

    Natalia N. Kudryavtseva,     Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Novosibirsk, Russia

    Vivek Kumar,     Bar Harbor, ME, United States

    Celeste Leung

    The Hospital for Sick Children, Toronto, ON, Canada

    University of Toronto, Toronto, ON, Canada

    Peter V. Lovell,     Oregon Health and Sciences University, Portland, OR, United States

    Anthony R. McIntosh,     Rotman Research Institute, Baycrest Centre, Toronto, ON, Canada

    Shannon McWeeney

    Oregon Health & Science University, Portland, OR, United States

    Oregon National Primate Research Center, Portland, OR, United States

    Adam Melgoza,     University of California San Francisco, San Francisco, CA, United States

    Claudio V. Mello,     Oregon Health and Sciences University, Portland, OR, United States

    Megan K. Mulligan,     The University of Tennessee Health Science Center, Memphis, TN, United States

    Michael A. Myre,     University of Massachusetts Lowell, Lowell, MA, United States

    Klotilda Narkaj,     University of Toronto, Toronto, ON, Canada

    Lydia Ng,     Allen Institute for Brain Science, Seattle, WA, United States

    Danton H. O'Day

    University of Toronto, Toronto, ON, Canada

    University of Toronto Mississauga, Mississauga, ON, Canada

    Denesa Oberbeck,     Oregon Health & Science University, Portland, OR, United States

    Ilse S. Pienaar

    Imperial College London, London, United Kingdom

    University of Sussex, Falmer, United Kingdom

    Robin Pierce,     Tilburg University, Tilburg, The Netherlands

    Firyal Ramzan,     University of Toronto Mississauga, Mississauga, ON, Canada

    Petra Ritter,     Charité – Universitätsmedizin Berlin & Berlin Institute of Health & Bernstein Center for Computational Neuroscience, Berlin, Germany

    Belén Sancristóbal,     Centre for Genomic Regulation (CRG), Barcelona, Spain

    David Schlessinger,     National Institute on Aging, National Institutes of Health, Baltimore, MD, United States

    Cornelia Schöne,     Department for BioMedical Research, University of Bern, Bern, Switzerland

    Robert Searles,     Oregon Health & Science University, Portland, OR, United States

    Puneet Sharma,     Imperial College London, London, United Kingdom

    Alexei A. Sharov,     National Institute on Aging, National Institutes of Health, Baltimore, MD, United States

    Dmitry A. Smagin,     Institute of Cytology and Genetics, Siberian Branch of Russian Academy of Sciences, Novosibirsk, Russia

    Ana Solodkin,     Anatomy & Neurobiology and Neurology, UC Irvine Health, Irvine, CA, USA

    Gilda Stefanelli,     University of Toronto Mississauga, Mississauga, ON, Canada

    Leon Stefanovski,     Charité – Universitätsmedizin Berlin & Berlin Institute of Health & Bernstein Center for Computational Neuroscience, Berlin, Germany

    Cindy Tao,     University of Toronto Mississauga, Mississauga, ON, Canada

    Ibrahim Tastekin,     Centre for Genomic Regulation (CRG), Barcelona, Spain

    Joe Z. Tsien

    Augusta University, Augusta, GA, United States

    Banna Biomedical Research Institute, Yunnan Academy of Science and Technology, Yunnan, China

    Nikki Walter

    Oregon Health & Science University, Portland, OR, United States

    Oregon National Primate Research Center, Portland, OR, United States

    Brandon J. Walters,     The Hospital for Sick Children, Toronto, ON, Canada

    J.T. Westwood,     University of Toronto, Mississauga, ON, Canada

    Robert W. Williams,     The University of Tennessee Health Science Center, Memphis, TN, United States

    Morgan Wirthlin

    Oregon Health and Sciences University, Portland, OR, United States

    Carnegie Mellon University, Pittsburgh, PA, United States

    Christina Zheng,     Oregon Health & Science University, Portland, OR, United States

    Joelle Zimmermann,     Rotman Research Institute, Baycrest Centre, Toronto, ON, Canada

    Iva B. Zovkic,     University of Toronto Mississauga, Mississauga, ON, Canada

    Preface

    The use of recombinant DNA technologies has revolutionized neuroscience research over the past 5  decades, created the field of biotechnology, and had a dramatic impact on translational research and medicine. Products resulting from the use of these technologies are found in laboratories, pharmacies, hospitals, and supermarkets all around the world. Discoveries made possible by recombinant DNA technologies propelled basic research, engaged the media, and fascinated lay people and scientists alike.

    The first edition of Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research provides an informative account of some of the most exciting molecular and recombinant DNA techniques of the 21st century as they are employed for the analysis of brain function and behavior. The information presented in this book is critical for clinicians, scientists, researchers, graduate students, and advanced undergraduate students, as well as professors interested in teaching related disciplines. The book provides a clear presentation of the frontiers of basic research, as well as translationally relevant techniques that are used by neurobehavioral geneticists.

    The first edition of Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research includes chapters on neuroinformatics, genetic and neurobehavioral databases and data mining, examples on the analysis of natural genetic variation, approaches to studying gene expression for understanding brain function and behavior, principles and applications of forward (mutagenesis) and reverse genetics (gene targeting), as well as discussions about the ethical use of animals in neurobehavioral genetics research. It is written by leading international experts of the field, and it is edited by a behavior geneticist who has been working in the biopharmaceutical and biotechnology research industry as well as in academic research for over 3  decades.

    Acknowledgments

    This book represents the dedicated and diligent work of the authors who wrote its chapters, but it would also not have been possible without the help from the numerous anonymous referees who provided their expert advice. I would also like to recognize the help and encouragement of the staff of the Elsevier editorial office, particularly Ms. Kristi Anderson and Dr. Natalie Farra. I thank Dr. Wim E. Crusio, my long-time colleague and friend, for initiating this book project and for contributing to and reviewing its first plans. Last, I must recognize the, most often, positive role my family played: my wife Julia and daughter Flora who put up with me working many late nights, and my son Mark, whose inquisitive questions always reminded me that the next generation of scientists are already knocking on the door. This book is for all of them and for everyone else who would like to learn about the genetics of behavior.

    Why Should We Use Genetics in the Analysis of Brain Function and Behavior? Practical and Theoretical Considerations

    Robert T. Gerlai,     University of Toronto Mississauga, Mississauga, ON, Canada

    Charles Darwin in his book On the Origin of Species (1859)¹ already realized that behavioral characteristics are not uniquely different from other phenotypical features in that they also must be subject to the forces of natural selection and evolve. Although the scientific field we call genetics today did not exist back then, and although the role of genes and structure of DNA were discovered only about a 100  years later, Darwin's theory of evolution implied that behavioral characteristics must have been under the influence of genes. This inherent assumption turned out to be correct and led to the development of a new field of science, behavior genetics. The birth of behavior genetics is usually dated to 1960, the year when the first book specifically focusing on questions addressed by this discipline was published.² In this book, the reader could find methods and applications of how one could study the effects of genes on a variety of behavioral traits, from perceptual processes to personality and temperament. The methods centered around what we now call quantitative genetics, a branch of classical genetics that is extended from Mendelian genetics to characteristics that cannot be classified into discrete categories but instead show continuous variation. Indeed, quantitative genetics was the main method of the behavioral geneticist for several years to come. It allowed scientists to conclude about genetic effects that underlie observed phenotypical differences. Quantitative genetics investigated the genetic architecture,³ i.e., it allowed one to estimate the source of genetic variance, e.g., whether it was due to additive genetic effects or to dominance-recessive allelic interaction-related genetic effects (intralocus interaction) or to epistatic effects (interlocus interaction). It allowed one to estimate the number of segregation units (likely genes with major effects), and it even allowed one to dissect phenotypical correlation and to estimate purely genetic and purely environmental components of correlation.⁴ What it did not allow one to do was to actually identify the gene, and to understand and/or to manipulate its biochemical or biologic function. That is, the gene itself remained an esoteric concept for behavioral geneticists.

    Molecular biology changed all this with the discovery of restriction endonucleases (e.g., Ref. 5), a discovery that revolutionized genetics, and biology in general. Restriction endonucleases allowed scientists to create recombinant DNA,⁶,⁷ and in doing so, to discover and manipulate genes and to probe their function. The past 5  decades have seen an amazing evolution of this technology. Genome editing in the 21st century is not a practice restricted to only a select few elite laboratories anymore. Service provider companies can custom design DNA, or genetically engineer laboratory organisms, for practically any research purpose according to the specifications of the biologist user. The biotechnology and biopharmaceutical research industry routinely utilize such technologies, and academic researchers all over the world employ and continuously develop increasingly powerful and novel methods, all falling under the category of the recombinant DNA approach. Behavior genetics has also benefited from this revolution.

    Recombinant DNA technologies now allow one to identify genes that show natural genetic variance in brain function, or to employ random mutagenesis and identify novel mutants with altered behavioral phenotypes and the genes that carry the mutation, an approach called forward genetics. Recombinant DNA methods now allow us to specifically target genes whose nucleotide sequence has been known but whose function in the brain is yet to be discovered, an approach called reverse genetics. They make it possible for us to check how our manipulation of behavior affected gene function, e.g., leading to transient or lasting changes in gene expression. But the story does not end there. Twenty-first century molecular biology now allows scientists to genetically manipulate the functioning of specific neuronal cell types, or circuits, giving unprecedented level of spatial resolution for the scientist. The genetic manipulation can also be conducted with extremely high temporal resolution: activation or deactivation of neurons can now be achieved at subsecond-level speed. Last, unlike in the past, genetic manipulation is now not restricted to a select few laboratory species, such as rodents, zebrafish, or the fruit fly, but it may be performed in a large variety of other animals too.

    Capturing this revolution is not an easy task. A novice, and perhaps even the expert, in this field may be overwhelmed by the sheer volume of publications, the large number of methods, and the even larger number of their applications. The current book cannot present this vast literature in a comprehensive way. But what it does attempt to do is to showcase some of the most powerful and most exciting developments of the fast-evolving field of behavior genetics, or as we now call it, neurobehavioral genetics. In doing so, the authors of our chapters, who are leading experts within their fields, will guide the readers, helping them make the best choices about how they could use a variety of molecular and classical genetic methods to study the effects of genes on brain function and behavior.

    From pharmacology to brain imaging, there are numerous elegant nongenetic methods in the tool set of the neuroscientist. Why should we bother with genetic methods or care about genetic effects on behavior and brain function? There are multiple reasons. One, because genes represent the most basic functional units of biologic organization, and thus manipulating them may allow us to understand how things are built and how things work in the organism. Two, manipulation of genes provides one of the most precise ways with which we can probe the functioning of the organism. We understand their structure, we have an amazing array of methods to modify them, and unlike environmental factors whose number is practically infinite, we have to deal with only 20–30 thousand of them. Three, comparison of nucleotide sequences across different species is perhaps the most precise way to discover similarities and differences among species. In other words, genetics allows us to translate findings from one species to another, including findings obtained with laboratory species to humans, which may lead to appropriate modeling and mechanistic understanding of human disorders. Last, the fourth reason is that without genetic effects, behavioral characteristics would not have evolved. Genes are the target of evolution, and behavior is the output of the brain, an organ that has been shaped by natural selection. Although a particular behavior and a gene that influences it are perhaps at the two most extreme levels of the multilayered biologic organization, evolutionary biology has taught us that they are inherently tied together. Thus, the preceding phenogenetic and phylogenetic arguments will likely keep the seemingly distant fields of psychology, neuroscience, and genetics married for a long time. The subsequent chapters of this book attest that this marriage has been a happy one, producing increasingly sophisticated methods and exciting discoveries.

    References

    1. Darwin C.R. On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. 1st ed. London: John Murray; 1859:502.

    2. Fuller J.L, Thompson W.R. Behavior Genetics. New York: John Wiley & Sons, Inc.; 1960:396.

    3. Crusio W.E, van Abeelen J.H. The genetic architecture of behavioural responses to novelty in mice. Heredity. 1986;56:55–63.

    4. Gerlai R, Crusio W.E. Organization of motor and posture patterns in paradise fish (Macropodus opercularis): environmental and genetic components of phenotypical correlation structures. Behav Genet. 1995;25:385–396.

    5. Roulland-Dussoix D, Boyer H.W. The Escherichia coli B restriction endonuclease. Biochim Biophys Acta. 1969;195:219–229.

    6. Cohen S.N, Chang A.C, Boyer H.W, Helling R.B. Construction of biologically functional bacterial plasmids in vitro. Proc Natl Acad Sci USA. 1973;70:3240–3244.

    7. Jackson D.A, Symons R.H, Berg P. Biochemical method for inserting new genetic information into DNA of Simian Virus 40: circular SV40 DNA molecules containing lambda phage genes and the galactose operon of Escherichia coliProc Natl Acad Sci USA. 1972;69:2904–2909.

    Section I

    Neuroinformatics, Computational Models and Data Analysis

    Outline

    Chapter 1. Neurological Biomarkers and Neuroinformatics: The Role of The Virtual Brain

    Chapter 2. Informatics for Interoperability of Molecular-Genetic and Neurobehavioral Databases

    Chapter 3. The Allen Brain Atlas: Toward Understanding Brain Behavior and Function Through Data Acquisition, Visualization, Analysis, and Integration

    Chapter 4. ExAtlas: Online Tool to Integrate Gene Expression and Gene Set Enrichment Analyses

    Chapter 5. Computational Models: How Do They Help to Understand Neurologic Diseases?

    Chapter 6. The Use of Recombinant Inbred Strains in Systems Genetics and Functional Analyses in Behavioral Pharmacology

    Chapter 1

    Neurological Biomarkers and Neuroinformatics

    The Role of The Virtual Brain

    Ana Solodkin¹, Joelle Zimmermann², Anthony R. McIntosh², Leon Stefanovski³, and Petra Ritter³     ¹Anatomy & Neurobiology and Neurology, UC Irvine Health, Irvine, CA, USA     ²Rotman Research Institute, Baycrest Centre, Toronto, ON, Canada     ³Charité – Universitätsmedizin Berlin & Berlin Institute of Health & Bernstein Center for Computational Neuroscience, Berlin, Germany

    Abstract

    An exciting advancement in the field of neuroscience has been the development of schemes to acquire data at the whole brain level, thus fostering a greater understanding of brain function in health and disease. Yet what we are clearly lacking are quantitative integrative tools to translate this understanding to the individual level. Here, we present a novel neuroinformatics platform, The Virtual Brain (TVB), to effectively model individualized brain activity, enabling the linkage of large-scale (macroscopic) brain dynamics with biophysical parameters at the meso- and microscopic levels. The remarkable consistency of changes in modeling parameters in our studies points toward feasibility to identify potential biomarkers associated with individual physiopathology in stroke and neurodegeneration. Previously, we have provided proof of concept demonstrating that TVB can create the basis for a more deliberate integration of computational biology and neuroscience into clinical approaches for elucidating cellular and molecular mechanisms of disease. The idea now is to create a library of model parameters associated with different neurologic diseases to further the development of individualized therapeutic interventions. In this chapter, we present an introduction to computational neuroscience and neuroinformatics, and in particular, we describe in detail the construction, usage, and relevance of the modeling approach with TVB. Furthermore, we discuss the role of biomarkers in medicine, their special requirements in the context of neurology, and how neuroinformatics can aid into their discovery. As there is a fair amount of technical terms, the reader is encouraged to refer to the glossary table listed at the end of the chapter.

    Keywords

    Alzheimer disease; Biomarkers; Brain dynamics; Precision medicine; Simulations; Stroke

    Adding Information to Brain Imaging Data Via Brain Network Modeling

    The neuroscience community is immersed in the collection of large datasets to provide greater sensitivity for understanding brain function and dysfunction. Such initiatives span normal function (Human Connectome Project), life-span including normal and abnormal function (UK Biobank),¹,² development (NIH Pediatric Database), brain disorders such as Alzheimer disease (ADNI),³ and psychiatry (RDoC: Research Domain Criteria Project). While these initiatives provide the necessary empirical foundation, what are lacking are the quantitative tools to relate these multiple datasets to reconstruct the brain and provide the link between these data and those from a single person. The technologic advances in brain imaging yield unprecedented information on structure and function, yet these data are severely underutilized for clinical decision-making.

    While in many cases group analyses can show statistically significant results between young and old or between healthy and diseased populations, strong underlying intersubject and intrasubject variability⁴ preclude accurate individual predictions. That is, MR imaging–derived markers are of small predictive value at an individual level.² While multiple investigations of potential imaging-based markers in preclinical studies are ongoing, clinical trials are hampered by their limited scope, as new innovative tools are needed to allow an exact assessment of the degree of disease pathology, including dynamic processes. In this chapter, we propose the advancement of current state-of-the-art approaches by using The Virtual Brain (TVB) simulation modeling platform, a novel method that explicitly infers network mechanisms at multiple spatial and temporal scales.

    TVB is an open-source whole-brain simulation platform that integrates empirical data of different modalities and spatiotemporal scales to construct comprehensive and biologically plausible models of brain network dynamics (http://www.thevirtualbrain.org/). Simplifying, TVB simulates mathematically driven brain activity on the three-dimensional map of an individual brain and its connections to fit empirically measured brain signals (from raw signals to electroencephalography [EEG], magnetoencephalography [MEG], and functional magnetic resonance imaging [fMRI]). By the systematic exploration of the relation between structural anatomic tracts and biophysical parameters producing the dynamics (spatiotemporal characterizations), TVB allows the construction of the individual's brain model within a unifying framework. One of the defining features of TVB is that it builds a unique Virtual Brain for each subject based on his/her own neuroimaging data within a plausible anatomic framework.⁵ This is of relevance since during aging and in various brain disorders, most studies have reported structural and functional changes focusing at the group level⁶–⁸ with highly variable intersubject relationships between structural abnormalities (lesions, etc.) and functional deficits. At the same time, similar deficits can be associated with different locations of damage where lesions involving large areas of the brain may result in only subtle deficits, whereas less extensive lesions may affect a number of cognitive domains. The basis for such interindividual variability is poorly understood because we do not understand the underlying functional brain dynamics associated with structural changes. TVB as a multiscale approach fills the gap by linking from cellular events to whole brain dynamics reflected in the modeling parameters, providing an opportunity to uncover basic mechanisms associated with brain function and dysfunction.

    Our preliminary work using this multiscale modeling approach has demonstrated the explanatory power of connectome-based, large-scale brain models⁹,¹⁰ and has facilitated the elucidation of the interrelationship between the structural and functional connectomes highlighting intersubject differences.⁹–¹¹ Furthermore (Fig. 1.1), we have also shown the algorithms to achieve optimal individual simulations (fit) of functional connectivity matrices based on their respective individual anatomic connectivity matrices.⁹ Finally, we have tested the dependency of functional connectomes and brain dynamics derived from damage to the structural connectome in stroke¹⁰–¹² and epilepsy.¹³,¹⁴ Because of the ability for TVB to generate personalized Virtual Brains via the simulation of functional signals, the model parameters associated with each healthy subject or patient can then be correlated to individual clinical phenotypes (e.g., motor function to cognition).

    TVB houses a library of local models, which catalog the biophysical parameters that produce different empirical brain states or network patterns.⁹,¹⁵ Table 1.1 lists the repertoire of neural mass models available in TVB and their state variables along biophysically relevant parameters. Because of the insight that TVB provides on these latent biophysical variables is invisible to noninvasive brain imaging techniques, it acts de facto as a computational microscope. TVB modeling is able to simulate a plethora of brain states and neural behaviors from raw data (e.g., neuronal firing) to brain signals detected as EEG or fMRI signals. Furthermore, these regimes are detected in condition-specific contexts (condition-specific regimes), enabling the model system to associate specific parameter profiles to condition-specific physiologic spatiotemporal behaviors. The overreaching goal is to provide proof of concept that TVB can in fact identify aging and disease biomarkers with predictive potential for disease diagnosis and progression based on underlying biophysical processes.

    Figure 1.1  The Virtual Brain discovery technology. The repertoire of tool chains, neuronal models, motif dictionaries, and educational support is continuously developed and advanced. This scheme shows the interrelationship of the components involved in the technologic development of The Virtual Brain platform.

    Table 1.1

    Finally, part of the approach of TVB is that it involves a broad community leading to an organized workflow of systematic knowledge inference going from the public (educational) domain¹⁶ to basic science research and clinical settings (Fig. 1.2).

    Bridging Scales Within The Virtual Brain

    Brain function derives from the complex interactions of the underlying microscopic structure, yet the description of a systems behavior, especially brain function from the bottom-up alone, is very limited in part due to redundancy of the basic elements (neuronal groups). In other words, it is not possible to deduct large-scale behavior based solely on aggregating independent small elements. The same can be said of the link between cellular and molecular levels. Classic neuroscience studies, focused at the level of single neurons and their physical interaction, do not use a bottom-up approach from the molecular level. That is, even when specific microstructure of axons, dendrites, glial components, and vesicles, for example, can be captured with very high resolution,¹⁸ this microcomplexity does not enable capturing information at the superordinate level, as the interaction of neurons and neuronal ensembles develop local functional patterns described by functional clusters and other nonrandom features.¹⁹ The new field of network science makes use of mathematical methods to describe these functional patterns.²⁰ In practical terms, functional modeling of individual human brains is achieved by defining and explicitly linking different network scales with some dimension reduction (to avoid excessive computational demands). In this context, choosing the right coarseness within each scale is an important step in developing multiscale models.²¹ For example, brain regions referred to as nodes in the modeling context are associated with the determination of their local dynamics via neural mass models, and their size has to be carefully determined. A sensitive description by Breakspear and Jirsa define a neural mass as a group of neurons with equal or similar in- and outgoing connections and a common function (similar to the definition determining the threshold of micro- and macrocolumns, i.e., populations up to 10⁴ or 10⁶ neurons).²² In our case, the TVB platform provides a wide range of multiscaling levels based on a variety of basic biophysical considerations implemented by access to a variety of local models (Table 1.1). A given model can in principle reproduce the neural population activity at the magnetic resonance imaging (MRI) resolution level with about 50,000 nodes, each maintaining approximately 5  million neurons. In addition, larger, functionally or anatomically driven parcellations are useful if the focus is on large-scale functional network features.⁹,²¹ TVB also introduces an intermediate mesoscopic scale, which pools local connectivity features, practically building a bridge between the macro- and the microscale. The mesoscale has no specific biological surrogate so far and is an operational construction to model microscopic population features in a large-scale network. Because of the fluid transitions between scales, it is necessary to define levels clearly, according to the respective case. But in general (Fig. 1.3), the following can be stated:

    Figure 1.2  Knowledge generation with personalized Virtual Brains. Multimodal imaging data are used to constrain biophysically grounded full brain models. Brain regions (regional map, parcellation) are represented by neural field models. Individualized brain network models can reveal the origin of interindividual differences in brain function and therefore are potential powerful tools for precision medicine. 

    With permission from Stefanovski L, Ghani A, McIntosh AR, Ritter P. Linking connectomics and dynamics in the human brain. e-Neuroforum. 2016;7(3):64–70.

    1. The microscale covers the irreducible elements of the network from the connectional point of view, i.e., single neurons and their membrane potentials. Functionally, it models individual action potentials and the microscopic circuits between single neurons. In the neural mass models of TVB, this scale is simplified as a neural mass; the circuits are represented, for instance, by the intrinsic interaction between the population of excitatory and inhibitory neurons.²²,²³

    2. The mesoscale is a surrogate intermediate scale that describes relevant population features, both in terms of function and structure. For example, it mathematically defines the average interaction between neighboring nodes by physical and functional laws²² or prominent modes of activity at the population level in a neural mass (e.g., in Stefanescu and Jirsa).²⁴

    3. The macroscale relates to the entire brain. It refers to emerging spatiotemporal neuronal activity modes resulting from local and global interactions of neuronal populations across the brain. With respect to structural features the macroscale refers to the structural connectome derived from diffusion tensor imaging (DTI) tractography and represents major white matter fiber tracts.

    There are no clear boundaries between the scales because of the fluid transitions from one scale to the next, suggesting that they are linked. Hence, TVB represents the microscopic level with the biophysically based equations of the neural mass models and its cellular and subcellular surrogate parameters where neural masses themselves represent the mesoscale (i.e., local field potentials and population firing rates of a network of several thousand if not million neurons). Likewise, distant large-scale connections (intracortical or projection) can be considered macroscopic, but regional tracts (e.g., between cortical layers or columns within a specific region) are mesoscopic. That is, scales are context dependent (Fig. 1.3), and the close integration of these scales is one of the most enduring features of TVB, as it fosters the determination of brain dynamics via faithful simulations of brain signals.

    Thus, TVB captures brain dynamics at the coarse scale of interacting neuronal populations (see evolution equation, Fig. 1.4), where local models provide a reduced or simplified description. These local models (Table 1.1) highlight different aspects of neuronal dynamics. Some are derived from models based on microscopically coupled neurons. For example, the Stefanescu-Jirsa 2D and 3D models are derived from networks of single-neuron FitzHugh-Nagumo and Hindmarsh-Rose⁷⁹ models, which in turn are derived from the famous single-neuron Hodgkin-Huxley model. In other words, TVB can be considered a top-down approach, where simulations are done at a coarser level at a whole brain level. In contrast, other modeling approaches, such as the Human Brain Project (https://www.humanbrainproject.eu/), are bottom-up, where simulations are done at subcellular and cellular levels to later infer activity at coarser scales. Independent of the approach, the ultimate goal based on noninvasive imaging is to ultimately understand the links between genes, proteins, and molecules to individual neuron physiology to neuronal population behaviors and eventually large-scale dynamics.

    Figure 1.3  Multiple scale dimensions in The Virtual Brain (TVB). Graphic representation of scales (structural and functional) associated with brain dynamics. The aspects colored with blue underground mark the scales formally represented in TVB. The triangle to the left is a reminder of the lack of clear border between scales. Morphologic level: Lower border of resolution is represented by columns as functional units. Specific cortical-brainstem tracts (corticospinal or medial lemniscus) are currently not established for TVB. Computational basement: As an artificial mathematic level, it is widely represented in TVB and (indirectly) models the whole brain. Structures below the neural masses have no direct representation in TVB. However, microscale features are essential in the development of TVB (e.g., in the mathematic reduction of microscopic models). Model output: In dependence of the used neural mass model, activity can be represented to the level of neuron populations in the form of subcellular features (as fitted parameters) and from the neuronal scale as spikes, action potentials, oscillations, etc. At the large-scale level, TVB has extensive options (monitors) representing activity (e.g., BOLD activity, EEG, MEG, and derived computations as the distribution of frequency powers). Empirical measurements: Because of its base the lower threshold of TVB is the resolution of MRI, as also for the network components.

    Figure 1.4  Integration of scales within The Virtual Brain (TVB) via evolution differential equations. (A) Graphic representation of global and local parameters based on the Stefanescu-Jirsa 3D (SJ3D). Each brain region (node seen as white circles) is represented by the SJ3D neural mass model that consists of inhibitory and excitatory population mean field dynamics that can infer mesoscopic population behaviors such as excitation-inhibition balance and firing rates, as well as intrinsic coupling strength between inhibitory and excitatory mean fields indicated here by parameters K11, 12, 21. Global parameters are derived from the associations (white lines) between the nodes. (B) Generic evolution equation of The Virtual Brain. TVB computes the evolution of a dynamic system—the brain—via differential equations. On the left side, it has a differential operator, and on the right, a vector of state variables provides a phase flow from where the system trajectory is integrated. One of the vectors (turquoise) represents the state variables for the local model (e.g., synaptic activity, firing rate, mean field potentials). Additionally, parameters derived from anatomic tracts map the current state of the system to the next (representing the interaction between nodes). Hence, these parameter symbolize global variables (purple and blue). An important global parameter (G, pink), a numeric scale reflecting the balance between local and global dynamics, varies according to the strength of anatomic connections. Additional elements in the equation (light and darker green) represent input simulation and noise to help to reproduce naturalistic biologic responses.

    Databases such as the UK Biobank or the Human Connectome Project—in addition, to brain imaging—provide detailed data on genetic traits, proteomics, and metabolomics along with information on lifestyle and environment. Hence, in the not so far future, it seems we will be able to link TVB to cellular and subcellular scales on the one hand and—similarly important—environmental and social factors on the other.

    Interestingly, at each level of description—from molecules to brain and to society—we are dealing with complex systems where rules and insights of nonlinear systems dynamics could apply. As each of these scales may have their peculiarities and specific rules, a new frontier is emerging to identify similarities and discrepancies among them, integrating them in complex manifolds.

    Pragmatics of Modeling in The Virtual Brain

    The sole empirical input to the TVB platform is the structural connectivity matrix derived from the individual subject's tractography. Based on this input, TVB simulates brain signals (e.g., field potentials) by integrating global dynamics with local (mesoscopic) models that determine the neuronal dynamics within brain regions. Following, EEG or fMRI blood oxygen level–dependent (BOLD) signals are derived from the generated field potentials. For a comprehensive list of the state variables and free parameters for the local models currently included in the TVB platform, see Table 1.1.

    Empirical data required for modeling with TVB (Fig. 1.5) are as follows:

    1. High-resolution T1-weighted anatomic images covering the whole brain. This allows for the anatomic location of brain regions. Regional brain parcellations are derived from this MR acquisition.⁹,²¹,²⁵

    2. Diffusion-weighted imaging (e.g., DTI) allows the determination of individual structural brain connectomes via tractography algorithms. The two metrics used for each tract are weight of connection and length of connection.

    3. Functional data, typically, falls into one of these three possibilities: resting state fMRI (rsfMRI)⁹,²⁶,²⁷ or intra- or extracranial electroencephalographic recordings.⁹,²⁸ Additionally, it is possible to use multiple modalities (e.g., simultaneous EEG-fMRI).²⁹

    Pragmatically, individual empirical structural and functional data can be directly extracted and formatted using an automated imaging data processing pipeline developed for TVB.²⁵ This TVB preprocessing pipeline²⁵ is already accelerated with parallel computing, and it is publicly accessible (http://www.nsgportal.org). The resulting structural and functional matrices from the pipeline can then be directly uploaded to the TVB platform.

    Figure 1.5  The Virtual Brain (TVB) modeling workflow. Graphic representation depicting the elements necessary for performing TVB modeling (represented by the central figure of brain nodes). (1) Empirical inputs include the individual cortical geometry and the respective structural connectome generated from T1-weighted magnetic resonance imaging acquisitions (allowing for brain parcellation and DTI tractography). (2) Selection of local (neural mass) model that will allow for the exploration and fitting of biophysical parameters. (3) The platform has an additional option of adding stimulation to any brain nodes. (4) Once parameter values are fitted, the simulated output is obtained. This can be a raw brain signal (local field potentials or neuronal firing rates) or EEG/MEG and BOLD signals. (5) Finally, the efficacy of the simulation is calculated by correlating it to the empirical signals (not depicted in the figure). 

    Modified from Kringelbach ML, McIntosh AR, Ritter P, Jirsa VK, Deco G. The rediscovery of slowness: exploring the timing of cognition. Trends Cogn Sci. 2015.

    In summary, large-scale brain network models in general, and TVB in particular, have few critical components: anatomic connectivity, neural (local) mass models, and brain lesions (patient populations). First, TVB uses the structural high-resolution T1-weighted MRI data to create a custom brain surface, and it uses diffusion-weighted MRI data to infer the anatomic connections among brain areas. Second, neural mass models describing the network nodes (brain regions) are used to determine the dynamic nature of the interactions in a biologically realistic manner. Third, an evolution equation integrating the connectivity between regions and the activity within regions calculates global brain activity over time, generating simulated brain signals for each region. Finally, acquisition of functional data (fMRI) permits testing the accuracy of the simulated signals generated by the model (Fig. 1.5).

    The TVB modeling workflow includes the following sequential steps performed for each individual subject (example based on the stroke work):

    1. Importing of the two subject-specific structural connectivity matrices (weights and lengths) into the TVB platform.

    2. Selection of the local neural mass model: The platform includes a variety of 12 neural population models. These can go from phenomenological models (e.g., Generic 2D, Epileptor) to those simulating local field potentials or synaptic activity (e.g., Wong-Wang, Stefanescu-Jirsa) and firing rates (e.g., Wilson-Cowan, Jansen-Rit). For example, in the stroke studies presented here, we use the Stefanescu-Jirsa 3D model,²⁴ as it permits characterization of the direct association between neural mass responses (field potentials) and rsfMRI.³⁰,³¹

    3. Parameter Space Estimation: Most of the modeling focuses on determining the optimal global and local parameters for each subject. This includes two steps. (a) Parameter space exploration: We use heat maps of global variance (mean variance of time series across all brain regions) to constrain the range of values for each model parameter. The desired range has high global variance flanked by bifurcation points²² (Fig. 1.6). (b) Parameter fitting: The derived optimal values result in the best fit between the empirical and simulated signals based on the following metrics: Global parameters: conduction velocity and global coupling; Local parameters: K12 (inhibitory on excitatory coupling), K21(excitatory on inhibitory coupling), and K11 (excitatory on excitatory coupling). The Stefanescu-Jirsa 3D model has several additional parameters (i.e., affecting ion channels, bursting strength, input current, and excitability¹⁰; see Table 1.1) that are kept at default values unless the simulated signals are not optimal.

    4. Stochastic network simulation: Based on the optimal values obtained, we generate field potentials. White noise with Gaussian amplitude (mean=0, standard deviation=1) is added to each node, and numeric integration of the system is performed using a stochastic Heun's method,³² with an integration step size of 0.0122ms.

    Figure 1.6  Parameter space mapping. These heat maps represent variance from a healthy subject (A) and a stroke patient (B). The plots show the global variance of the model's dynamics as a function of changing parameter values for conduction velocity (y axis) and global coupling (x axis). A noteworthy feature is that topography on the stroke case clearly differs from the parameter map associated with healthy controls. These differences are most noticeably in the areas of higher variance (dark red) flanked by bifurcation points (white arrows). These values are essential for parameter exploration and the final parameter fitting.

    5. Determining BOLD signals: These forward models can then be derived from the field potentials using a hemodynamic response function implemented with a gamma kernel²³ with the same duration and sampling rate (TR) as the empirical rsfMRI acquisition.

    6. Assessing the accuracy of the simulated time series: We compare the empirical and simulated BOLD time series in terms of amplitude, frequency, and phase. Amplitude is calculated from the highest and lowest peaks present in the time series across all regions. Frequency is computed via Fast Fourier transform using Matlab's fft function with a sampling frequency of 0.5Hz to determine the range, profile, and peak frequencies.⁹ We assess phase by comparing the functional connectivity matrices of the simulated and empirical time series via Pearson's correlation.

    7. Excluding overfitting and ensuring model identifiability: The number of free parameters in the model does raise concerns of overfitting and identifiability (i.e., whether there is convergence to a single optimal solution). To this end, we examine convergence to a global extreme, and then, with a fivefold cross-validation, we exclude overfitting. For each subset (fold), the functional (predicted) data are randomly divided into an 80% sample as training segment and a 20% sample as testing segment.

    Note that some of these computations—both in the preprocessing pipeline and in the TVB modeling per se—are very demanding in terms of processor time and memory demands, requiring novel computational infrastructures³³ and high-performance computing.³⁴ Fortunately, more efficient approaches to perform some of the computational tasks are in development.³⁵

    Modeling With The Virtual Brain in the Clinical Context

    Stroke

    Focal damage to the brain results in disruption of a distributed network of connections (white matter tracts) initiated in the damaged region(s), making network analyses particularly useful in studying the widespread effect of brain injury³⁶–³⁸ (for a review, see Ref. 39). Our approach focuses on using methods from network neuroscience to contribute to a personalized data-driven approach to neurological care (i.e., precision medicine⁴⁰). We hope that by building computational models integrating data from clinics and from basic research, we can develop a mechanistic understanding of the disease process,⁴⁰,⁴¹ and in the context of the recent push to collect large datasets, to develop an approach to making useful diagnostic and therapeutic inferences from such complex data.

    As previously mentioned, multiscale brain network models are gaining increasing attention⁴²–⁴⁴ because they produce biologically relevant interpretations and operate via simulation of brain activity as a mean to understand its properties. That is, the formal link between different observational levels is established by numeric simulations resulting from the integration of local and global dynamics,⁴⁵ where the biophysical parameters at the local level describe the properties of small populations of neurons, and at global level, they represent biological mechanisms governing interactions between brain regions.

    Relationship Between Recovery After Stroke and Modeled Biophysical Parameters

    In our previous work,¹⁰ we sought to determine changes in brain dynamics after stroke by modeling local and global parameters. In this study, motor performance was assessed before, after, and long after (6–12  months) a therapeutic intervention for hand motor skill, yet the modeling was solely done based on the imaging before therapy. Based on individual structural connectomes, we showed that patients with stroke demonstrated a consistent reduction in conduction velocity, increased local dynamics, and local excitability (via a decrease in inhibitory coupling). Importantly, we found a negative relationship between local excitation and motor recovery and a positive correlation between local dynamics and motor recovery (Fig. 1.7). Thus, our TVB modeling revealed a disrupted poststroke system favoring excitation-over-inhibition and local-over-global dynamics, consistent with existing literature on stroke mechanisms.⁴⁶–⁴⁸ This work is an important step demonstrating the potential of TVB to determine individualized biophysical changes associated with stroke recovery. In a second study,¹¹ we aimed to relate these results to existing literature and subjected each structural connectome to graph theoretic analysis to determine the relationship between modeled global coupling and the organization of anatomic tracts.⁴⁹,⁵⁰ For this, we calculated degree centrality, betweenness centrality, and global efficiency.⁵¹ Linear regression analysis demonstrated that our TVB parameter global coupling was negatively correlated with global efficiency (P  =  .038), but not with degree centrality or betweenness centrality. In other words, the larger influence of local dynamics seen through changes in global coupling parameter is closely associated with a decreased efficiency of the system in stroke.

    The Virtual Therapy

    In an additional study,¹² we piloted the question of Virtual Therapy (VT), elicited within TVB by changing specific simulation parameters toward control values. In this study, we hypothesized that the normalization of parameter values to those found in healthy controls would result in the reestablishment of healthier brain dynamics even in the presence of a damaged brain. After performing VT (i.e., normalizing the parameter to control levels), 17 of 20 stroke subjects showed an increased correlation of resting state functional connectivity with the resting state functional connectivity of an average of healthy control participants (Fig. 1.8). The percent gain ranged from 5.0% to 85.7%, and the combination of parameters that produced the largest improvement in brain dynamics included global coupling and local excitatory on inhibitory coupling. In other words, the change in these parameters produced normalization of the balance between global and local dynamics and excessive excitation.

    Figure 1.7  The Virtual Stroke Brain. The figure highlights salient aspects of TVB modeling to generate the Virtual Stroke Patient Brain. First, the modeling involves the determination of the effects of structural damage to brain dynamics (represented by the axial T1-weighted image on the left). Second, the assessment involves local and whole brain dynamics (center figure); finally, the scatter plots to the right represent the close relationship between local dynamics (balance of excitatory-inhibitory coupling) with motor function after therapy (upper graph) and 1  year after therapy (lower graph). The higher the dynamic change was, the worse was the motor recovery.

    Neurodegeneration: Alzheimer Disease

    Alzheimer disease is a complex degenerative process of the brain, which begins in ventromedial cortical limbic regions before it spreads to the rest of the brain.⁵²,⁵³ The long, chronic degeneration with a loss of a critical number of neurons leads to Alzheimer dementia (AD), which is characterized by an ongoing loss of memory and higher cognitive functions. Remarkably, one of the hallmark neuropathologic markers of AD (neurofibrillary tangles) disrupts axonal integrity, suggesting AD is a classic disconnection syndrome. Indeed, AD is marked by cognitive dysfunction emerging from altered brain network activity.⁵⁴–⁵⁶ Interestingly, alterations across similar brain networks are also implicated in cognitive impairments seen in Parkinson disease (PD),⁵⁷ where cognitive dysfunction can appear in the final stage of the idiopathic Parkinson syndrome.⁵⁸ Determining differences and similarities between AD and PD poses a significant challenge because of overlapping affected brain networks captured by changes in large-scale brain network studies.⁵⁹ To develop new therapeutic approaches to dementia, an initial question to resolve is to determine individuals vulnerable to develop dementia in preclinical stages before degeneration is extensive.

    Figure 1.8  The Virtual Therapy. (A) Example of the Virtual Therapy (VT) implementation for one stroke case. Parameters were normalized from the baseline value to the healthy control average. This process was repeated for each patient. (B) Frequency histogram showing the number of stroke cases stratified by percent of change in the correlation of functional connectomes from baseline after the VT in relation to healthy controls. In all but three cases, the functional connectivity matrices after the therapy were closer to those in healthy controls. Changing the parameter values in the stroke subjects to mimic those in controls resulted in healthier brain dynamics even when the damage did not change.

    As TVB is a connectivity-based simulator targeting neural populations⁹ underscored by structural and functional connectivity changes (Matthews et al., 2013; Sun et al., 2014)⁸⁰,⁸¹, along with local changes in brain dynamics, we recently initialized a study to determine the specific global and local biophysical parameters associated with the generation of the AD clinical phenotype. That is, to understand the functional pathology in AD, we are currently investigating biophysical TVB parameters associated with cognitive profiles from healthy controls, subjects with mild cognitive impairment (MCI), and Alzheimer disease.

    For this, structural connectomes derived from 124 subjects (16 AD, 35 MCI, and 73 healthy controls) from the Sydney Memory and Aging Study were included in the modeling. Based on individual anatomic connections, personalized Virtual Brain models were constructed using the Reduced Wong-Wang neuronal mass model included in the TVB repertoire. The model successfully simulated rsfMRI time series for each subject. To determine the relationship between optimized model parameters and clinical phenotype, neuropsychologic cognitive measures were integrated in a single score and correlated with parameter measures also integrated in a single score (Fig. 1.9).

    Our preliminary results not only hint to consistent changes in biophysical parameters, but they also show a consistent positive relationship with the clinical phenotype. Based on this last feature, we are currently expanding the analysis to determine the statistical membership of MCI subjects to either the AD or the healthy control groups, as this is a crucial step to determine converters to AD and, hence, pave the way in the development of novel therapeutic interventions.

    Figure 1.9  The Virtual Alzheimer Brain. The figure highlights salient aspects of The Virtual Brain modeling to generate the Virtual Alzheimer Patient Brain. First, the modeling involves the determination of the effects of structural degeneration to brain dynamics (represented by the coronal T1-weighted image on the left along two microphotographs of neuritic plaques and neurofibrillary tangles). Second, the assessment involves local and whole brain dynamics (center figure); finally, the scatter plot to the right represents the relationship between brain dynamics (local and global combined) with combined cognitive scores. Note the relationship between these two metrics for healthy controls (black

    Enjoying the preview?
    Page 1 of 1