OHBMEducation-2022

Repo for the 2022 OHBM Education 1/2 day course entitled: "How to Write a Re-Executable Publication"

When: June 19th, 1pm-5pm local time

Where: Glasgow, Scotland, Scottish Event Campus (SEC), Alsh Room

Objective

The goal of this 1/2 day (4 hour) course is to have the students DO a re-executable publication.

Instructors (for the credit)

• Julie Bates - University of assachusetts Chan Medical School
• Dorota Jarecka - MIT
• Peer Herholz - McGill and ???
• Stephan Heunis - Institute of Neuroscience and Medicine, Forschungszentrum Jülich
• Neda Jahanshad - Universty of Southern California
• Stephen Strother

Orginizers (for the blame)

• David Kennedy - University of assachusetts Chan Medical School
• JB Poline - McGill

Prerequsites

• A computer with internet and browser
• A GitHub Account (Need one, go here!)

Tentative Schedule

1:00 - 1:05 (5 minutes) Introduction to the course (DNK/JBP) - Link to slides - Link to video

1:05 - 1:20 (15 minutes) What is a ReproPub and why would we want one? (JFB) - Link to slides - Link to video

1:20 - 1:35 (15 Minutes) Introduction to the publication you are going to do (NJ) - Link to slides - Link to video

• ENIGMA has some Parkinson’s Disease (PD) results
• You have some ‘new’ PD data
• Do you see what ENIGMA did?

1:35 - 1:50 (15 minutes) What are containers, in general (DJ) - Link to slides - Link to video

1:50 - 2:05 (15 minutes) How to make a specific “simple container” (containing FSL) (PH) - Link to slides - Link to video

2:05 - 2:20 (15 minutes) Brief intro to DataLad (aka DataLad does it all...) (SH) - Link to slides - Link to video

• installing data
• running containers
• publishing results

2:20 - 4:30 “Just do it” - Exercise

• Here are your command lines
• Here is the JupyterHub
• Do it, lots of ReproStaff around to help answer questions or problems

2:30 - 2:45 Break

4:30 - 4:45 (15 minutes) Meta and mega analysis of your results (NJ+DNK) - Link to slides - Link to video

4:45 - 5:00 (15 minutes) Recap and Summary (SS) - Link to slides - Link to video

Approach Overview

What do we mean by 'do a re-executable' publication?

• DataLad install a particular dataset - DataLad@NITRC
• DataLad containers-run a particular container (that generates some derived images and results in NIDM)
• DataLad "Publish" the resulting dataset
• pynidm query the results, and run a specific statistical test - PyNIDM@NITRC
• "Publish" the NIDM results to the ReproLake