Towards reproducible computational biology

An introductory tutorial

This is an introductory tutorial for working in a reproducible manner in bioinformatics/genomics and related fields of study. You will learn how to analyse some next-generation sequencing (NGS) data. The idea here is not to facilitate the best possible analysis of this data or to use the best tools available, but rather to learn some tools that are available to us for making our analysis as reproducible as possible. The data you will be using is real research data, albeit down-sampled to make the analyses finish in a reasonable time.

Currently, Dr. Sebastian Schmeier is teaching this material at Massey University in Auckland, New Zealand.

More information about other bioinformatics material, tutorials, and our research can be found on the webpages of the Schmeier Group (https://sschmeier.com).

Note

A PDF-version of this tutorial can be downloaded here.

Slides from a recent introductory seminar to this tutorial given at ResBaz Auckland 2018 can be found below:

Contents

• 1. Introduction
  • 1.1. Prerequisites
  • 1.2. Learning outcomes
  • 1.3. The data we will be using
  • 1.4. The analysis workflow
• 2. Working reproducible
  • 2.1. What is the problem?
  • 2.2. How to tackle this problem?
  • 2.3. Background reading on reproducibility
• 3. Tool and package management
  • 3.1. Installing the Conda package manager
    • 3.1.1. Update .bashrc and .zshrc config-files
    • 3.1.2. Installing conda channels to make tools available
  • 3.2. Using conda to search and install tools
  • 3.3. Create isolated environments
  • 3.4. General Conda commands
• 4. Creating analysis workflows
  • 4.1. What is a workflow management system?
  • 4.2. What is Snakemake?
  • 4.3. Setup
    • 4.3.1. Install Snakemake
    • 4.3.2. Download data
  • 4.4. The analysis without a workflow management system
    • 4.4.1. Data QC
    • 4.4.2. Read mapping
    • 4.4.3. Feature counting
    • 4.4.4. Saving tool version information
    • 4.4.5. Summary
  • 4.5. Using a workflow management system
    • 4.5.1. Snakemake
    • 4.5.2. Creating rules
    • 4.5.3. Error logging and benchmarking
    • 4.5.4. Integrating package management
    • 4.5.5. Running Snakemake
    • 4.5.6. Visualising the workflow graph
    • 4.5.7. Building the remaining rules
  • 4.6. Making your work available
• 5. Containerization
  • 5.1. What is containerization?
  • 5.2. What does it accomplish for us?
  • 5.3. Using a Singularity container
  • 5.4. Building your own Singularity container locally
  • 5.5. Building a container on Singularity Hub
  • 5.6. Using a container in our workflow
  • 5.7. Using one container for the whole workflow
  • 5.8. Ready made containers
  • 5.9. Combining containers with conda-based package management
  • 5.10. Background reading on containers
• 6. Downloads
  • 6.1. Tools
  • 6.2. Data
  • 6.3. Tutorial