Program

SLAS2016 Short Courses

Study Design and Statistical Analysis for High Throughput Screening (HTS) Experiments

As data analysis methods for high-throughput screening continue to evolve, methodological research has shown that the reproducibility and validity of HTS screens can be greatly improved by modern design and statistical methods.

Approximately 1/3 of the course will cover basic study design principles (e.g., randomization, internal validity, avoiding confounding variables) and introductory statistical principles (e.g., systematic versus random error, data visualization, inferential versus exploratory data analysis, false positive versus false negative errors). The remaining time will be spent applying these principles to HTS experiments, covering both primary and secondary (validation) screens.

Recent advances from the published methodological literature will be covered from an applied perspective, both in lecture format and by live data analysis sessions conducted by the instructor. The R statistical software will be used to provide examples of data analysis sessions with emphasis on the ideas and the statistical output rather than on the coding mechanics. R scripts will be provided in a format suitable for post-workshop use by participants or by bioinformatics colleagues. Participants will not run R code during the workshop.

Who Should Attend:

The target audience are industry and academia individuals with an interest in HTS data analysis, including graduate students, post-doctoral fellows, scientists, bioinformaticians, screening facility personnel, and laboratory heads. Attendees should have hands-on or strong conceptual experience with HTS.

How You Will Benefit From This Course:

The aim is to give participants a working knowledge of how to use statistical principles to help plan HTS experiments and to analyze the data to improve reproducibility and validity. References to accessible methodological literature will be provided, including books that can be examined at the workshop. Material will be provided to participants to allow them to apply knowledge gained in the workshop to their own data sets. Methods for biochemical and cell-based assays will be addressed, although note that RNAi screens and image-based High-Content Screens will not be covered in this workshop.

Course Topics:

Introduction to Study Design and Statistical Principles

  1. Reproducibility Crisis and the Need for Better Study Design and Analysis
  2. Statistical Concepts and Terminology
    • Logic of Statistical Tests
    • What is a p-value?
    • Assumptions of Statistical Tests
    • Data Transformation
    • Data Visualization — Advantages and Caveats
  3. Study Design
    • The Importance of Design and Minimizing Confounds
    • Minimizing False Positives and False Negatives
    • Replication

High Throughput Screening Study Design and Analysis

  1. Primary Screen Systematic Error (Bias) Detection
  2. Primary Screen Normalization and Study Design Methods
  3. Primary Screen Statistical Testing
  4. Partial Replication
  5. Secondary (Validation) Screen Normalization and Study Design Methods

Instructor:

Robert Nadon, Ph.D.,

  • Associate Professor, Department of Human Genetics, Principal Investigator, McGill University
  • Principal Investigator, McGill University and Genome Quebec Innovation Centre

Robert Nadon is an Associate Professor in Department of Human Genetics of McGill University, a Principal Investigator at the McGill University and Genome Quebec Innovation Centre in Montreal, Canada, and an Associate Editor for BMC Bioinformatics. Dr. Nadon's research focusses on analysis of high throughput biological data from various sources, notably gene expression microarrays, genome-wide mRNA translation, and high-throughput screening. He has published on these topics in a wide range of technical and life science journals, including Nature Biotechnology, Journal of Biomolecular Screening, PNAS, Bioinformatics, BMC Bioinformatics, and Trends in Genetics. He consults regularly with industry and has given various high-throughput data analysis workshops for life science audiences.

Partial References:

General
HTS