2020-Q1

March 31st 2020

Multiple updates were added to the literature import and tagging module, as well as more flexibility in the risk of bias/study evaluation section by allowing study-level overrides to specific questions. We've also expanded the visualization libraries being used, making it easier to make new visualizations in future releases. Stability was improved by adding a database fixture for easier testing, updating all packages to recent versions, and adding full windows support to the developer environment.

We also released both a Python and R-based HAWC client. This allows for developers, data scientists, and scientists to interact with both reading and writing data to HAWC via code instead of manually in the user-interface. Expect a dramatic expansion of the API in future releases.

Full changelist below:

• General

  • Replace chemspider with the EPA chemicals dashboard
  • Add the ability to select language for NOAEL/NOEL/NEL, LOAEL/LOEL/LEL for assessments
  • Updated all packages to latest version for stability and easier development in Windows
  • Update django to version 2.2 LTS

• Literature

  • Fix imports from HERO using new HERO API
  • Addition validation when importing from HERO to ensure all selected references for import can be imported into HAWC
  • Allow faster querying of PubMed using the new PubMed API key
  • Customizable literature tagtrees - can change which nodes are displayed and filter references by selected tags
  • Preserve UI settings for expanded/collapsed tags when tagging literature
  • Increase maximum PubMed query size to 10,000
  • Update reference detail layout; add related searches/imports for references
  • Show full author list instead of subset; full-author list is now also searchable when finding references
  • Add new histogram for literature overview by year
  • Added new API endpoints for interacting with literature module
    • /lit/api/assessment/{id}/tags/ - view all tags for assessment
    • /lit/api/assessment/{id}/reference-ids/ - view references for assessment and related PubMed/HERO ids
    • /lit/api/assessment/{id}/reference-tags/
      • GET: view references and associated hawc tags for each reference
      • POST: append/replace existing reference-tag relationships with new relationships
    • /lit/api/search/
      • POST: added new create HERO import API for importing references from HERO

• Risk of bias app

  • Add new risk of bias overrides feature
    • multiple scores can be selected for a single metric/study combination
    • labels can provide context to which data the score are relevant for
    • extracted bioassay or epi components can be associated with overrides
    • updated all visualizations and data views to present override results in addition to default

• Visualization

  • Add conditional formatting to data pivot reference-bars
    • commonly used to show if a trend-test is significant
  • Wrap-text automatically for long tag-node names in literature trees

• Client

  • Added new Python-based HAWC client for reading/writing data to HAWC
  • Added new R-based HAWC client for reading/writing data to HAWC

• API

  • enabled token-based authorization to allow standard automated tooling to access the HAWC API
  • Created new tabular style exports - for tabular REST API, can multiple standard output formats by specifying format=xlsx, or {csv, xlsx, json, html}

• Testing

  • Add a database test-fixture to dramatically expand unit-testing capabilities
  • Updated test to use fixture instead of creating objects in test-setup
  • Created a new test for asynchronous celery task execution in admin
  • Added tests for phantomjs (svg -> png) conversion stability

• Developer environment

  • Support Windows as a developer environment for HAWC
  • Updated developer documentation
  • Add new commands to anonymize developer database
  • Simplified developer environment setup to make the settings/local.py file optional
  • Expanded available /admin/ views
  • Update docker deployments to use recent container versions
  • Added new API endpoints for rendering plotly visualizations

2019-Q4: hawc merge

We've merged the teamhero/hawc branch into the main branch, and have done a large amount of package updating and refactoring to make the developer environment easier to use and ready to take advantage of recent python improvements including automated formatting and type-checking.

• Overall changes
  • black + flake8 linting and formatting enforcement
  • Restructed python code into a true python package
  • Update to webpack3 including linting and formatting enforcement
  • Added CI and automated testing
  • Added utility command-line scripts to capture all data pivot visuals SVG and PNG
  • Added server level hawc flavors which allow for different behavior across controls {EPA, PRIME}
    - update admin pages and default views
• Assessment-level changes
  • Allow nomenclature options at the assessment level (NOEL/NOAEL and LOEL/LOAEL; "Risk of bias" vs "Study evaluation")
    - added more content cleanup options
• studies
  • added the ability to "lock" studies so they cannot be edited, unless a project-manager unlocks
  • added new options for COI reporting
    - animal bioassay:
    - animal husbandry and diet moved from experiment to animal
    - text displays as NOAEL or NOEL depending on assessment-specific settings
    - animal group endpoint table is now sortable by multiple fields
    - expanded endpoint copy form to allow more richer sorting and filtering
    - epi:
    - Study populations can belong to multiple countries instead of just one
    - Epi exposures can now report multiple "central tendency" descriptions of exposure instead of one
• risk of bias
  • Allow alternate names for Risk of Bias / Study evaluation
  • Different score-choices available based on hawc "flavor" (epa vs prime)
  • An single overall study confidence metric can be created per reference, if needed
  • Added new selector to toggle if metric is required for invitro data
  • Add assessment-level instructions for conducting evaluation
  • risk of bias/study evaluation scores can be shown/hidden at the assessment level
  • donut plots have domain text rotated
  • metrics can have shorter names to be presented in tables and visuals
  • risk of bias information added to data-pivots for sorting/filtering/displaying

2018-Q3

2018 Q3 release.