testingGIN/minutes/2019-1202_NFDIworkshop_magdeburg.md

Presentation

Magdeburg about to write a grant. What topic to talk about in the afternoon.

Thomas Wachtler: NFDI-neuro

(LMU): usual presentation. Q: what about BRAIN initiative'?

Michael Denker: reproducible analysis of neuroal activities

tools for electrophysiology and modelling.

scenario data reuse

• jack wants the data and the code and be able to reproduce the analysis, make new ones and use the same analysis on a different dataset. -> repeatability -> replicability (new team) -> reproducibility (new experiment setup)

odML, Neo

Making collaboration possible. metadata standard, computer readable neo: abstraction from raw data

Rescience journal

standardised description of specific implementations: elephant library

high potential, but still quite scary to me.

how to implement these tools in the lab?

• having someone using the tool
• get ERC/phd students to learn (by doing)

how to write it up ?

1. incentives for the data producer

- Who is doing what with my data
- Did someone attempted to use the data ?

Is there a way to implement such a collaboration.

Data curationb after the fact is difficult.

2. Costs (time to invest) to choose/learn the tools

MRI data (Michael Hanke)

http://datalad.org

• imaging data is personal data
• data is too big
• no metadata standards

challenges

• not open, but FAIR
• reusability and interoperability is way to costly
• no incentives to make data accessible

what data is important

• there is some investment, where to put the ressources
• answer depends on who you ask (researcher, data steward, tech, librarian,...)

datalad

• workflow for data input to output archiving.
• capable of making transition to new standards.

handbook.datalad.org

sharing analysis, not data

• if we get the data in a standard format, we could actually do that.

today tool disappear: sould become a inconvenience, nothing more

• self contained

#Thorsten TRR135 and nowa

what we do

• DMP and evaluation of workflows
• software for workflow assistance
• training

Tools

RDM planning tool (RDMO), heesen box (seafile) Gitlab with quality checks Research data repository (Dspace)

Challenges: variability

• location
• organizational structures
• research methods
• data volume
• standards usage and awareness (and existence)

Rene Bernard, neurocure

6 institutions -Berlin based, 45% is Charité 25 PIs, clinical and basic research 2 partner institution (BIH, DIFE-potsdam)

value and open science module

Success and enabler

• ELN (600 active users now)
• reward open data
• QUEST toolbox
• Qumula scalable data storage (2020) - linked to ELN

...

Ideas

• get one responsible person in the lab - someone easier to access.

• go iterative, small steps first.

• good practive example

NFDI

• linking tools

• sharing platform

• RDMP = blablabla

• knowledge repository

• financial ressources

• DFG will not tell it loud, but SFB funding depends on data management.

• more focussed discussion

• see the appliaction as objective:

• Get things done, have outputs (strengthen the application)

• dialog with field and sfb, update guidelines for DFG application

• more collaboration : entire field exploiting the existing ressources.

• working group implementations

Tools:

• tutorial based infromation (in context)

openbrainconsent workflow: collect, check what works, make one file that works