permucn 0.1.0

Permutation-based test for copy-number and binary trait association

permucn

permucn is a command-line tool for testing whether trait transitions (0->1, optionally 1->0) are associated with gene-family copy-number evolution from CAFE outputs. In binary mode, you can choose either permutation-based testing or Fisher's exact test with Tarone screening.

Install

Requirements: Python >=3.12

pip install "permucn @ git+https://github.com/mkrg01/permucn.git"

If published on PyPI:

pip install permucn

Optional plot support:

pip install "permucn[plots] @ git+https://github.com/mkrg01/permucn.git"

Quick Start

1. Fetch sample data (works after pip install):

permucn get-test-data --out-dir permucn_test_data

2. Run permucn:

permucn \
  --cafe-dir permucn_test_data/toy_example/cafe_output \
  --trait-tsv permucn_test_data/toy_example/species_trait.tsv \
  --no-include-trait-loss \
  --n-perm-initial 20 \
  --n-perm-refine 50 \
  --seed 7 \
  --out-prefix results/toy_binary

3. Check outputs:

• results/toy_binary.family_results.tsv
• results/toy_binary.run_metadata.json
• results/toy_binary.top_hits.tsv
• results/toy_binary.top_pvalues.tsv (default: top 100 by p-value)

If at least one tested family has p-values, these are also written:

• results/toy_binary.pvalue_hist.tsv
• results/toy_binary.qq.tsv

For a larger sample dataset:

permucn get-test-data --dataset polar_fish --out-dir permucn_test_data

permucn \
  --cafe-dir permucn_test_data/polar_fish/cafe_output \
  --trait-tsv permucn_test_data/polar_fish/species_trait.tsv \
  --jobs 4 \
  --perm-cache results/perm_cache.json.gz \
  --out-prefix results/polar_fish

Required Inputs

--cafe-dir must include:

• Gamma_change.tab (required)
• Gamma_asr.tre (required)
• Gamma_branch_probabilities.tab (required only with --cafe-significant-only)
• Gamma_family_results.txt (optional)

--trait-tsv must be a TSV with:

• one species column (species, taxon, name, etc.; first column fallback)
• one binary trait column (0/1)

Important rules:

• Species names in trait TSV must match tree tip names.
• If multiple binary trait columns exist, specify --trait-column.
• --cafe-significant-only is valid only in binary mode.
• rate mode requires strictly positive non-root branch lengths.

Common Commands

Default binary mode:

permucn --cafe-dir <cafe_output_dir> --trait-tsv <trait.tsv> --out-prefix results/binary

Binary mode with CAFE-significant events only:

permucn --cafe-dir <cafe_output_dir> --trait-tsv <trait.tsv> --cafe-significant-only --out-prefix results/binary_sig

Binary mode with Fisher + Tarone:

permucn \
  --cafe-dir <cafe_output_dir> \
  --trait-tsv <trait.tsv> \
  --binary-test fisher-tarone \
  --out-prefix results/binary_fisher

Rate mode:

permucn --cafe-dir <cafe_output_dir> --trait-tsv <trait.tsv> --mode rate --out-prefix results/rate

Reproducibility and Performance

• --seed: reproducible permutations
• --jobs: parallelism (1 sequential, 0 auto CPU)
• --perm-cache: reuse permutations across runs (.json / .json.gz)
• --pvalue-top-n: size of *.top_pvalues.tsv (0 disables)

Documentation

• wiki/Getting-Started.md
• wiki/CLI-Reference.md
• wiki/Input-Format.md
• wiki/Output-Interpretation.md
• wiki/Algorithm-Notes.md
• wiki/FAQ.md

License

MIT (LICENSE)