picard is an R package for building and running reproducible, pipeline-based RWE studies using OHDSI tools. It provides a standardized study repository structure, manifest-driven input management, and execution workflows from development through production.
What picard Helps You Do
picard helps analysts manage the full lifecycle of an RWE pipeline, from study bootstrap through execution and dissemination.
1. Launch a Standard Study Repository (Ulysses)
picard initializes a consistent repository structure and project scaffolding via the Ulysses setup flow.
- Defines study metadata and contributors
- Configures one or more database blocks
- Creates standard folders, config files, and starter scripts
- Supports Git and reproducibility setup during initialization
2. Maintain Execution Settings
picard centralizes DBMS connection and execution configuration through ExecutionSettings.
- Builds execution settings directly or from config blocks
- Separates schema settings from credential storage
- Supports environment-specific runs and cohort table naming conventions
3. Manage Inputs with Manifest Tools
picard uses manifest objects to register and track cohort and concept set inputs as durable study assets.
- Imports from ATLAS or adds programmatic definitions
- Registers custom SQL and derived cohorts
- Tracks metadata, hashes, dependencies, and status
- Supports mid-cycle maintenance (check, update, delete, reset)
4. Create Pipeline Files and Scripts
picard provides file-generation helpers to reduce manual setup and enforce project conventions.
- Creates task scripts and reusable source utility files
- Creates SQL templates for analysis logic
- Creates pre-pipeline builder scripts for cohorts and concept sets
- Creates dissemination scripts for downstream reporting workflows
5. Execute and Monitor the Pipeline
picard supports consistent pipeline execution across study tasks and database targets.
- Runs cohort generation and task pipelines with execution settings
- Tracks task execution state, rerun logic, and summaries
- Supports structured workflow from development to production runs
Install
# install.packages("remotes")
remotes::install_github("OHDSI/Picard")
library(picard)Quick Start (Minimal)
library(picard)
sm <- makeStudyMeta(
studyTitle = "OMOP Characterization Study",
therapeuticArea = "Endocrinology",
studyType = "Characterization",
contributors = list(
setContributor(
name = "Jane Doe",
email = "jane.doe@institution.org",
role = "developer"
)
)
)
db <- makeBlock(
configBlockName = "my_cdm",
cdmDatabaseSchema = "omop_cdm_schema",
cohortTable = "study_cohorts",
databaseName = "my_database_v1",
databaseLabel = "Primary CDM"
)
ulySt <- makeUlyssesStudySettings(
repoName = "my_study_project",
repoFolder = "~/studies",
studyMeta = sm,
dbConnectionBlocks = list(db)
)
ulySt$initUlyssesRepo(verbose = TRUE, openProject = FALSE)For detailed usage examples and end-to-end workflows, see the package site and vignettes.