testingGIN/minutes/2019-04-09-sourcedata

TODO
====

- learn about texture
- prepare meeting may
- work with other tools (gin, hubox,...)?


talk with Thomas Lemberger
==========================

Goals
-----

Initial (ultimate) goal was to link figures with data, early in the scientific process.

To achieve this goal, the idea was to make some reasonable data curation step, that would be a continous curation, i.e.:

- cost effective, at embo below 10% of costs
- level of the figure: add some structured metadata to each figure
- aimed at molecular and cellular biology, expandable ??

How it works?
------------

- entities are found (text data mining)
- entity type defined (7-8 possibilities)
- entity role defined (test or control)
- link to ontologies


Achievement:
-----------

- 30 000 papers curated
- test and control conditions tagged differently

Future:
-------

- collaboration with texture: http://substance.io/texture/
    - produces jats formats
    - creation of sourcedata metadata before publication
- need to restrict ontologies used at the lab level (choose form a small list?)
- data kept outside sourcedata (link to it)
- demo should be ready in May


Julien's 2 cent
--------------

- use the material list as source of information to ease the link to ontologies (RRID to gene names for example)
- need a search engine which makes sense and which starts with more information than a keyword. From what people published, and other info (protocol used, material in stock,...) one could build a context to search data/figures that make more sense. This could be discussed with Peter Kraker: https://www.go-fair.org/implementation-networks/overview/discovery/

Other discussion
----------------

- convince people by stating what they would need from other researchers
- Thomas to be in Berlin on May the 8th
- external dvt, few people
- ideas for extra funding ?!
- RDA very abstract, Force11 very idealists
- develop bioschemas.org