Modern Statistics for Modern Biology

Description

If you are a biologist and want to get the best out of the powerful methods of modern computational statistics, this is your book. You can visualize and analyze your own data, apply unsupervised and supervised learning, integrate datasets, apply hypothesis testing, and make publication-quality figures using the power of R/Bioconductor and ggplot2. This book will teach you 'cooking from scratch', from raw data to beautiful illuminating output, as you learn to write your own scripts in the R language and to use advanced statistics packages from CRAN and Bioconductor. It covers a broad range of basic and advanced topics important in the analysis of high-throughput biological data, including principal component analysis and multidimensional scaling, clustering, multiple testing, unsupervised and supervised learning, resampling, the pitfalls of experimental design, and power simulations using Monte Carlo, and it even reaches networks, trees, spatial statistics, image data, and microbial ecology. Using a minimum of mathematical notation, it builds understanding from well-chosen examples, simulation, visualization, and above all hands-on interaction with data and code.

Susan Holmes is Professor of Statistics at Stanford University, California. She specializes in exploring and visualizing multidomain biological data, using computational statistics to draw inferences in microbiology, immunology and cancer biology. She has published over 100 research papers, and has been a key developer of software for the multivariate analyses of complex heterogeneous data. She was the Breiman Lecturer at NIPS 2016, has been named a Fields Institute fellow, and is currently a fellow at the Center for the Advances Study of the Behavioral Sciences. Wolfgang Huber is Research Group Leader and Senior Scientist at the European Molecular Biological Laboratory, where he develops computational methods for new biotechnologies and applies them to biological discovery. He has published over 150 research papers in functional genomics, cancer and statistical methods. He is a founding member of the open-source bioinformatics software collaboration Bioconductor and has co-authored two books on Bioconductor.

'This is a gorgeous book, both visually and intellectually, superbly suited for anyone who wants to learn the nuts and bolts of modern computational biology. It can also be a practical, hands-on starting point for life scientists and students who want to break out of 'canned packages' into the more versatile world of R coding. Much richer than the typical statistics textbook, it covers a wide range of topics in machine learning and image processing. The chapter on making high-quality graphics is alone worth the price of the book.' William H. Press, University of Texas, Austin 'The book is a timely, comprehensive and practical reference for anyone working with modern quantitative biotechnologies. It can be read at multiple levels. For scientists with a statistics background, it is a thorough review of key methods for design and analysis of high-throughput experiments. For life scientists with a limited exposure to statistics, it offers a series of examples with relevant data and R code. Avoiding buzzwords and hype, the book advocates appropriate statistical practice for reproducible research. I expect it to be as influential for the life sciences community as Modern Applied Statistics with S, by Venables and Ripley or Introduction to Statistical Learning, by James, Witten, Hastie and Tibshirani are for applied statistics.' Olga Vitek, Northeastern University, Boston 'Navigating rich data to arrive at sensible insight requires confidence in our biological understanding, informatic ability, statistical sophistication, and skills at effective communication. Fortunately the wisdom and effort of the worldwide research community has been distilled into accessible and rich collections of R and Bioconductor software packages. Holmes and Huber provide a comprehensive guide to navigating modern statistical methods for working with complex, large, and nuanced biological data. The presentation provides a firm conceptual foundation coupled with worked practical examples, extended analysis, and refined discussion of practical and theoretical challenges facing the modern practitioner. This book provides us with the confidence and tools necessary for the analysis and comprehension of modern biological data using modern statistical methods.' Martin Morgan, Roswell Park Comprehensive Cancer Center, leader of the Bioconductor project 'Holmes and Huber take an integrated approach to presenting the key statistical concepts and methods needed for the analysis of biological data. Specifically, they do a wonderful job of building these foundations in the context of modern computational tools, genuine scientific questions, and real-world datasets. The code showcases many of the newest features of R and its dynamic package ecosystem, such as using ggplot2 for visualization and dplyr for data manipulation.' Jenny Bryan, RStudio and University of British Columbia '... the book is extremely readable and engaging, it explains complicated concepts in simple terms, and uses illuminating graphics and examples. Any researcher who wants to learn or teach up-to-date statistics to biologists will find this an essential volume for modern teaching of modern statistics to modern biologists.' Noa Pinter-Wollman, The Quarterly Review of Biology