23. Reproducible workflows
Griffin Chure gave a guest lecture on reproducible workflows and using version control software to facilitate them. He inspired us to do this because “Everything you do as a scientist is for other people to use.”
Guidelines for reproducible research
He gave five guidelines to keep in mind when conducting research.
Reproducibility starts from project inception, not completion.
Draft the layout of your analysis, even if you don’t know how it will change.
Document as you go along, not when your colleague wants to use your code.
Separate your code by what it does, not when you’d use it.
Sunlight (UV) is the best disinfectant. Work in the open unless legally forbidden.
Links to resources
Griffin showed an interesting study on computational reproducibility by Victoria Stodden and coworkers, which you can download here.
He presented an example of how version control and careful organization of his code and protocols was used to do reproducible research to study the energetics of molecular adaptation in transcriptional regulation. You can access the repository here The permanent link to the archived version of the repository used in the publication can be found here.
He also presented a current project on futile protein expression, which you can access (and watch progress!) here.
He also provided a template repository for how he structures his research that you can clone and use for your own work. You can access it here.
Griffin encourages you to contact him at gchure at stanford dot edu if you would like to discuss or have questions.