OMOP Athena → DuckDB

athena2duckdb converts an extracted OMOP vocabulary download from OHDSI Athena into a ready-to-query DuckDB file with typed CDM tables, primary keys, and automatic row-count validation. The loader ingests the standard OMOP vocab files (CONCEPT.csv, VOCABULARY.csv, etc.) and skips auxiliary exports like CONCEPT_CPT4.csv or README.txt by default.

Features

• Discovers standard OMOP vocabulary files such as CONCEPT.csv, VOCABULARY.csv, CONCEPT_RELATIONSHIP.csv, and more.
• Streams each file into DuckDB using read_csv with quoting/escaping disabled, preventing parse failures caused by embedded quotes or backslashes, while the CLI shows a live progress bar per table.
• Loads recognised vocab files into typed tables (INTEGER, DATE, VARCHAR) that match the CDM DDL with primary keys already enforced (secondary indexes can be added later if needed).
• Always performs row-count verification to ensure the database matches source files.

Installation (local)

uv sync

or build/install directly from the project root:

uv build
pip install dist/athena2duckdb-*.whl

CLI Usage

uv run athena2duckdb /path/to/athena-export -o vocab.duckdb --verbose

Arguments:

Flag	Description
input_dir	Directory that contains the Athena CSV/TSV files.
-o, --out	Output DuckDB database file (default vocab.duckdb).
--sep	Field delimiter (default tab).
--encoding	Source file encoding (default UTF-8).
--threads	Number of DuckDB threads to use.
--schema	DuckDB schema name for created tables (default main).
--overwrite	Replace an existing DuckDB file if present.
--verbose	Emit INFO-level logs during the load.

Example

uv run athena2duckdb data/ -o vocab.duckdb

Sample output:

Loaded 10 tables into vocab.duckdb.
Tables: concept, concept_ancestor, concept_class, concept_relationship, concept_synonym,
domain, drug_strength, source_to_concept_map, relationship, vocabulary
OK        table=concept                  csv_rows=93,547 table_rows=93,547
...

Programmatic API

from pathlib import Path
from athena2duckdb import CSVOptions, load_vocab_dir, verify_row_counts

summary = load_vocab_dir(Path("data"), Path("vocab.duckdb"), schema="cdm")
results = verify_row_counts(summary.db_path, summary.vocab_files, schema=summary.schema)

Testing

uv run pytest

License

This project is licensed under the MIT License. See LICENSE.