Skip to content

CLAUDE.md

Latest commit

 

History

History
83 lines (56 loc) · 3.28 KB

CLAUDE.md

File metadata and controls

83 lines (56 loc) · 3.28 KB
Raw

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

DataQualityDashboard is an R package (part of OHDSI/HADES) that assesses data quality of OMOP CDM databases. It runs ~4,000 parameterized checks across three categories (Plausibility, Conformance, Completeness) and produces results viewable via a Shiny dashboard.

Common Commands

# Build and check
devtools::build()
devtools::check()
R CMD check --as-cran --no-manual .

# Run all tests
devtools::test()

# Run a single test file
testthat::test_file("tests/testthat/test-executeDqChecks.R")

# Run tests matching a pattern
testthat::test_file("tests/testthat/test-executeDqChecks.R", filter = "Execute a single DQ check")

# Lint
lintr::lint_package()

# Install local dev version
devtools::install()

Architecture

Execution Flow

executeDqChecks() → SQL generation (SqlRender) → DB execution (DatabaseConnector)
  → evaluateThresholds() → calculateNotApplicableStatus() → summarizeResults()
  → writeJsonResultsTo() / writeDBResultsTo()
  → viewDqDashboard() (Shiny)

Check Definition System

Checks are defined in CSV files at inst/csv/ — one set per CDM version (5.2, 5.3, 5.4). Each CSV row defines a parameterized check (table-level, field-level, or concept-level). The ~24 check types correspond to SQL templates in inst/sql/sql_server/. SqlRender translates these to the target database dialect at runtime.

Public API (6 exported functions)

  • executeDqChecks() — main entry point; runs checks and returns/writes results
  • viewDqDashboard() — launches Shiny app from a results JSON file
  • listDqChecks() — list available checks without executing
  • reEvaluateThresholds() — re-score existing results with new thresholds
  • writeJsonResultsToJson() / writeJsonResultsToCsv() — output format conversion
  • convertJsonResultsFileCase() — convert result keys between snake_case and camelCase

Key Source Files

File Purpose
R/executeDqChecks.R Main orchestration logic
R/evaluateThresholds.R Pass/fail/warning scoring
R/calculateNotApplicableStatus.R Determines when checks don't apply
R/recordResult.R Stores individual check results
R/summarizeResults.R Aggregates counts/percentages
R/writeJsonResultsTo.R / R/writeDBResultsTo.R Output writers
inst/shinyApps/ Dashboard Shiny app

Testing

Most tests use Eunomia (synthetic OMOP CDM data in-memory via DuckDB) — no external database needed. Connection setup is in tests/testthat/setup.R. Many tests use withCallingHandlers(..., warning = ...) to muffle expected check-name warnings.

Tests also exist to test the package on other DBMS; these connect to remotely hosted warehouses using creds stored in .Renviron.

Snapshot tests live in tests/testthat/_snaps/.

CDM Version Support

Checks are versioned: inst/csv/OMOP_CDMv5.2_Check_Descriptions.csv, 5.3, 5.4. When adding or modifying checks, changes typically need to be applied across all relevant CDM version CSVs and their corresponding SQL templates.

Linting

.lintr excludes sql/, output/, docs/, and inst/doc/. Run lintr::lint_package() before submitting PRs.