Example usage:

1. Convert the wide- first-event endpoint file (R13v2.0) to a zstd-compressed CSV:

pv finngen_R13_endpoint_2.0.txt.gz \
  | zcat \
  | xsv fmt -d"\t" \
  | zstd -T0 -q -o ~/data/r13_mine/data/finngen_R13_endpoint_2.0.csv.zst

2. Run the Endpcorr init:

uv run --python 3.13 --with numpy,pandas,pyarrow \
  main.py init \
  -e <(zstdcat ~/data/r13_mine/data/finngen_R13_endpoint_2.0.csv.zst) \
  -o ~/data/r13.0_results/correlations/intermediate__r13__2026-07-03.parquet \
  -d ',' \
  -x NA

3. Run the Endpcorr compute:

uv run --python 3.13 --with numpy,pandas,pyarrow \
  main.py compute \
  -i ~/data/r13.0_results/correlations/intermediate__r13__2026-07-03.parquet \
  -o ~/data/r13.0_results/correlations/

4. Run the Endpcorr csv:

uv run --python 3.13 --with numpy,pandas,pyarrow \
  main.py csv \
  -i ~/data/r13.0_results/correlations/ \
  -o ~/data/r13.0_results/correlations/correlations__r13__2025-07-03.csv