Skip to content

February 20, 2018, Tuesday

Jump to bottom
Liya Wang edited this page Feb 26, 2018 · 12 revisions
  • Ideas
    • For de.sciapps.org, switch back to Agave for file browsing
      • User can see all of their data (results are still saved in sci_data)
      • Create the sci_data folder for users if not existing
        • Tony already handled it? Will it be a delay in permission setting?
      • This will allow us to create the Atmosphere image
        • Remove requirements on https?
        • Add app management? Like add category, add app with category, etc so that app json can be automatically created
    • For sciapps.org, switch back to TACC and DS
      • Still use maizecode user to browse sci_data and submit jobs to TACC
      • Allow us to release workflow prototypes and continuously modify them
      • Need to figure out visualization for GWAS, annotation since results are now live at DS
        • For Manhattan plot, maybe upload the data
        • For browser, use DE links
    • For de2.sciapps.org, the only version that supports local jobs
      • Upgrade to MySQL or connect to Halcott?
      • Pending on completion of de.sciapps.org and sciapps.org
    • Tutorials
      • Modify the tutorial part for visualization
      • Add tutorial for the community version
  • Issues
    • Final fix in delay of saving running workflow by saving immediately
    • For a completed workflow, why do we need to keep refreshing for status? This causes a problem when checking for a file by scrolling down the diagram. The refreshing always move the focus to the top part of the diagram if the diagram is very tall
  • Improvements
    • maxCardinity
    • Building arbitrary workflows
    • Load app form over diagram?
    • Avoid staging too much data at one time, especially with replicates? Maybe we should only submit 2nd (or 3rd) job when 1st (or both 1st and 2nd) job is running so that we are not staging more than 1 (or 2) jobs at the same time
    • Allow 'job name'?
    • Run as maizecode user? since we can't afford concurrent staging? But might be blocked by Agave so not a good idea
  • Atmo
    • Remove iRODS component (also on de.sciapps.org, we can create the sci_data folder in the backend)
  • API
    • Swagger + MySQL
    • Focus on the workflow level with curls for list, status, submit, cancel, delete
    • Command line SDK?
    • Use metadata to check pair-end reads for job submission?
Clone this wiki locally