• Dr Julien Colomb,data curator
• @j_colomb
• berlin open science meetup organiser
• steering committee OSMOOC
• worked for Uni Jena RDM helpdesk for 8 month: https://rdmpromotion.rbind.io (a hugo based website)
• freelance teacher at A2P
• ex-neurobiologist, work with Y. Winter a couple of time

• requested by funders
• increase visibility/citations
• defence against accusation of fraud, build trust
• foster collaboration
• allow preliminary and meta-analysis
• accelerates research !
• well organised, speed up the research process ?!

• no time ?
• zip it + push it to zenodo + add authors = 30 minutes
• Fear people will find errors?
• If there are errors, do you really want no one to find them ?
• Ashamed of the data status ?
• Design your work for it to be shared (next time)

Reviewer ask for all figures with points instead of boxplots.

• create 11 boxplot images with points
• concatenate them into the 5 figures
• upload them online, privately
• get a link giving access for the reviewer
• https://figshare.com/s/7681d9bea3bd62220fd5

• How long ?

• took 10 minutes

Need to filter all data with score pretraining above 0.66

• 10 experiments to deal with
• filter data
• make statistics
• create boxplot (with sample size and stars for statistical significant differences)

• How long ?

• took 30 minutes

1. set variables and data path
2. raw data loaded
   • putative filtering
3. analysis performed
4. statistics produced
5. graphs produced via a reusable code or a function
   • Same look for all figures
   • using a different function -> different visualisation

• 3 copies in 2 locations

Task = analyse home cage scan data using machine learning algorithms. Run it through the data produced so far.

Data from Prof. Steele, no data management:

• only got uninteresting hourly summaries
• grouping not clear

Data from the AOCF, with data management plan:

• I find the data in minutes :)
• Raw data was saved, it is a text file :)
• Get animal history and genotype easily.
• File naming rules exist, :) but
  • changed over time or were not followed completely :(
  • special character were used :(

Solution: forget about Steele data, move the AOCF data, rename some files (Prüß -> Pruess), create an index (semi-automatically), add information about the animals in the index: https://www.overleaf.com/12764720rmhzqbpvtwxd

Now I can load the data.

• Folder organisation
• File naming
• Data index

The stupid future self of me will take care of that shit!

• Your future self is talking shit about you.

• Plan all ahead:
  • Folder organisation and file naming
  • Data format along the data flow
  • Where will it be published ?
• Apply and document your own rules:
  • Implementation is key
  • Check regularly you are still following the rules
  • document, document, document

https://www.bihealth.org/en/notices/workshops-on-open-data-175-and-206/