Enabling the reusability of scientific data: Experiences with designing an open access infrastructure for sharing datasets

1

• Enabling the reusability of scientific data
  Experiences with designing an open access
  infrastructure for sharing datasets

Simon J. Coles EPSRC National Crystallography
Service School of Chemistry University of
Southampton
2
Data the Publication Problem
2,000,000
25,000,000
450,000
3
A Different Approach to Data Publication?
Underlying data
Intellect Interpretation
4
Requirements

• Capture of all digital data and information
  generated during the course of an experiment
• Data validation
• Adding value
• Archival system for data with attached
  bibliographic and chemical metadata
• Automatic report generation
• Schema and protocols for publication and
  dissemination of a dataset

5
Open Access Crystal Structure Archive
ecrystals.chem.soton.ac.uk
6
Access to the Underlying Data
7
Publicising Content
8
Harvesting, Linking and Aggregating
9
Usability Quality Uniformity of data

• Different laboratories, practices instruments
  present a heterogeneous body of data
• Publish according to IUCr ratified schema
• To support publication according to this schema a
  toolbox add-on to the archive has been developed
• Toolbox requires 2 mandatory files only is
  capable of performing file format conversions and
  generate value added files

10
Usability Ease of Deposition Metadata Quality

• Minimal number of manual metadata entries many
  can be hardwired into the system
• Deposition guidelines initially prepared by
  students to provide impartial feedback
• Full documentation and in-line help/examples
• Restrained lists, e.g. Keywords
• Data deposited automatically by toolbox
• Automated generation of metadata for report and
  OAI interface

11
Usability Data Validation

• Peer review removed from self deposit publication
• Simple checks for consistency made by the toolbox
• Checks for crystallographic integrity made
  through a web service (IUCr, CHECKCIF)
• Introduction of data editor for the archive a
  deposition must be signed-off by a recognised
  professional before going live
• Quality indicators automatically taken from
  dataset and presented in HTML jump-off page

12
Usability Identifiers

• URL of deposited dataset provides an identifier
• Persistent only if the Institutional support
  model is accepted / adopted
• Signed-up to an agency to register metadata
  relating to datasets with a DOI
• Pay registry to ensure that DOI always resolves
  to associated dataset (10cents to register 1cent
  per annum to maintain)
• InChI chemical identifier - a unique text
  descriptor for a molecule

13
Usability Dissemination Aggregation

• OAI metadata schema ratified by IUCr chemical
  community
• OAI covers bibliographic terms must introduce
  chemical terms
• Both library and subject specific aggregators
  satisfied
• Chemical linking InChI, chemical classifications
  and restricted keywords list

14
Usability Endorsement

• Feedback during development from technical
  publishing arm of IUCr
• Designed for automatic incorporation into CSD
  (global database operated by CCDC)
• Accepted by Executive Committee of IUCr
• Reuse of data achieved in collaboration with
  Leverhulme Centre for Molecular Informatics

15
Usability Community Uptake

• Southampton archive about to publish routinely
  via the archive
• Five crystallography laboratories in UK agreed to
  adopt philosophy, install and populate archives
• CCDC will harvest required data from all archives
• IUCr will harvest and curate all data
• Develop aggregator services in collaboration with
  IUCr

16
Usability The Next Challenges

• Full acceptance by chemical community
• Validation worries
• Curation worries
• The requirement for as many peer reviewed
  publications as possible (despite quality)
• Full acceptance by wider chemistry publishing
  community
• Loss of control over underlying data
• Faith in Open Archives replacing experimental
  descriptions in articles
• Development of fully functional aggregator
  services