Data Management

"A biologist prefers to share toothbrush than to share gene names" former EU Bioinformatics Head

Slide:

https://github.com/rasmusagren/repres_slides/raw/master/data_management.pdf

Intro

Why managing research data?

* Make life easier
* Make it reproducible, avoid accusation of fraud

Why data open access? (Will be enforced soon)

* Transparency
* Enable new methods & reduce duplication
* To be reusable

Data management plans (will be enforced soon)

* Collection
* Organization
* Documentation
* Storage
* Policies & Sharing etc

Structuring Data

Poor data organization --> slower resarch

Suggestion

* Folder raw data with dates
* Code is kept separated from data
* Use version control for code
* Make scratch directory for experiment
* make Readme in each directory
* Protect key files
* Use filename scheme (use number or date first on files)
* Use non-proprietary formats  --> e.g. csv instead of xls

Example folders in project directory:

* doc/
* data/
* code/
* notebooks/
* logs/
* scratch/
* results/
* others/

Notebooks! VERY Important:

* Dated Entries
* Verbose
* Link to data and code
* Image
* Observation, Cocnclusions and ideas
* SOmething that doesn't work, so we don't need to redo it

Data Publishing and Re-use

* Long term storage
* Persistent Identifier like DOI
* Discoverable

• BIG ISSUE: Data persistence is a problem. Broken Links, unavailable resource
• Deposition database is available in elixir-europe.org