draft

[david-mears-2]

• Initial pre-processing script
• Add data inputs (examples) and output
• Update pre-processing script to make use of tidyr and dplyr
• Use a custom list of in-scope variants, per-STAVE release.
• Don't filter down STAVE data at all
• Extract R script from pre-processing draft md; use scripts dir for data inputs
• Use command line args for R script
• Potential fix for pull request finding
• mv .gitkeep
• Small script refactors
• Add README.md
• Use cli package for messages and errors in process_stave.R
• Add script to generate example model output parquet files
• Add metadata for model outputs
• Rename column variant_string to variant to match survey data parquet
• wip wip wip wip
• Make versions soft-coded in metadata endpoint
• Make prevalences parquets include all ancestor region ids, not just region id of most granular level
• Implement prevalences endpoint
• Update example model outputs generator to (1) go back to 1970 more often and (2) not run out of memory when running
• Add instrumentation for timing requests
• Fix survey data requests scoped by admin0
• Response optimizations: columnar format and round high-precision floats
• Rows for survey data, columns for prevalences
• Fix keys/names of rounded columns
• Extract grout region metadata fetch into a startup script
• Extract some consts/types
• Extract some validators
• Extract DuckDB connection
• Extract querying logic
• Refactor querying logic
• Move error handler call to correct place
• Finish implementing query param validators and refactor them
• Tidy comments; extract todos into youtrack tickets
• Reword variable names
• Make node types package match canonical node version
• Apply tsconfig to scripts ts files too
• Tests - iteration 0
• Tests - iteration 1 (refactor fixture generation)
• Remove fixture validation