Talk:canSAS-2012

2012-07-28 Notes of Discussions

• Data storage formats
  • sasCIF
  • newer standard for publishing bio SAXS data in IUCr journals (doesn't specify format, but includes data validation)
  • canSAS1d
  • NeXus
  • IUCr Diffraction Data Deposition Working Group proposal, SAS Commission response to proposal, & whether proposal would impact SAS data standardization
  • Considerations for the design of a canSAS-multid format
    • purpose of the format?
    • ease of use?
    • ease of maintainability?
    • user support?
    • what content?
    • can it handle 'big data'?
    • can it cope with sparse data?
    • extensibility?
    • well-defined?
    • ensure it gets used!

• NeXus main page
  • How to find links to Definitions and Examples? Make better main page links.
    • NAPI examples don't work (Ron Ghosh). What is the problem? Fix?
    • Needs more help oriented for data consumers
  • NXsas: a definition for raw SAS data
  • NXiqproc: general definition, might work for reduced SAS or at least be considered a starting point

• IUCr World Database of Crystallographers
  • Is this list contemporary?
    • Who are the consumers of this information?
    • Are standard terms codified?
    • Search for "small-angle" or "small angle" is not as useful as "SAXS" or "SANS"
    • SAS does not consider itself under "Crystallography"

PDB data deposition task force

(Information from email chat with Jill Trewhella)

Task force remit: Bio-molecular Small-Angle Scattering Data and Interpretation; Data Deposition and Model Archiving

Task Force Membership: Helen Berman, Wayne Hendrickson, Gerard Kleywegt, Guy Montelione, Mamuro Sato, Torsten Schwede, Dmitri Svergun, John Tainer, Jill Trewhella

Preliminary Recommendations:

• Need a standard dictionary for definitions of terms for data collection and for managing a data repository. X-rays and neutrons.
• Need a data repository that holds standard data set, is searchable, internationally accessible for download.
• Bead and atomistic models based on SAS data should be archived in a searchable data base and (given that the underlying data are not atomic resolution) there be specific information in regard to the quality/reliability/uniqueness of the result. Criteria for these assessments need to be defined; issues chi squared, error propagation, sensible stereo-chemistry requirement is argued and needs yet to be agreed.
• Need a meeting of key people from different task forces to come together to define what the pdb should be archiving.

We also discussed what might be in the PDB and what might be in a parallel "hybrid model" data base that is distinct from the PDB, but with a common interface. By this means it would be easy to communicate that the hybrid models are not based purely on data that can provide atomistic (or something close to it) information.

ADDITIONAL COMMENTS:

• While a database of data is desirable from the PDB perspective, it is not clear who should be the steward
• Data should include information about sample and its history/preparation etc., and treatment of background (and potentially storing background data as well)