Key words: teaching, mcgill, open data science, reproducibiliy, neuroscience, neuroimaging
Neuroscientists rely increasingly on data accessible online and on data science procedures for their investigations. Data science now offers a key set of tools and methods to efficiently analyse, visualize and interpret neuroscience data. Concurrently, there is a growing concern in the life sciences that many results produced are difficult or even impossible to reproduce. This is referred to as the reproducibility crisis, which concerns most of biomedical fields [F. Collin].
This first week of the computational neuroscience seminar series is bringing together software and analytical tools and methods. It will teach students how to best use the fundamentals of data science in their daily work to produce reproducible results. We will take examples in neuroimaging or imaging genetics, and see how to use computational tools, statistical and machine learning techniques, and collaborative and open science methodology to generate results that are statistically solid and computationally reproducible. While a large part of this course is language agnostic, we will teach and use python throughout the course. Students will start to work on projects that they will continue during the following computational neuroscience seminar weeks.
The course will require that you have basic programming experience and one or more undergraduate course(s) in statistical analysis (or equivalent experience), but it will be aimed at life scientists (neurologists, psychiatrists, pyschologists, neuroscientists) who wish to improve their research practices, or students who want an introduction to data science with examples in neuroscience and neuroimaging.
The first week, outlined below, can be conceptually divided into three sections:
- Part I: Introduction and Motivation
- Part II: Reproducibility and Data Management Tools
- Part III: Data analysis: concept and tools
- 09:00 - Course Introduction
- 10:00 - Epistemology and lessons from the past
- 12:00 - Lunch
- 13:00 - Installation time and troubleshooting.
- 14:00 - Git and Github
- 17:00 - Dismissal
- 09:00 - Python for Data Analysis
- 12:00 - Lunch
- 13:00 - Using and building Containers
- 16:30 - Assessment 1
- 17:00 - Dismissal
- 18:00 - Optional social event
- 09:00 - Standards for project management and organization
- 12:00 - Lunch
- 13:00 - Guest Lecture on Binder
- 14:00 - High-Performance Computing and Compute Canada
- 17:00 - Dismissal
- 09:00 - Introductory statistics
- 12:00 - Lunch
- 13:00 - Guest Lecture on estimation of connectivity
- 14:00 - Classical machine learning
- 17:00 - Dismissal
- 09:00 - Introduction to Deep Learning
- 12:00 - Lunch
- 13:00 - Multivariate statistics and matrix factorizations
- 16:00 - Assessment 2
- 17:00 - Dismissal