User Feedback 2018-07-09

[dmccloskey]

Present

Lars, Pasquale, Douglas

Feedback

Bottlenecks

• large learning curve
  • multiple .csv files to setup
  • easy to copy and past wrong .csv
• no automatic checks of user input
  • valid and consistent sample names between .csv files
  • valid and matching component_names between .csv files
• no log, error message, or user interface to easily alert the user to what is wrong
• need for a clear protocol for filling out the .csv input files
  • Google doc?
• plotting takes time (but provides incredible confidence during the analysis)
• distinguish between peak and no peak (false positives)
  • Time consuming to understand why something was filtered out
  • Flagged samples are time consuming to filter out manually
  • need to remove from summaryConcentration but keep in featureSummary with "used_ = false"

Other

• need to be able to run the workflow completely independently
• Docker stability

Benefits

• automation
• ease of calling the script