ecomp 0.1.0

Lossless evolutionary-aware multiple sequence alignment compressor

Docs · Report Bug · Request Feature

Evolution-informed lossless compression of multiple-sequence alignments (MSAs).

---

Installation

From PyPI (recommended for users):

# create and activate a virtual environment
python -m venv venv
source venv/bin/activate

# install ecomp
pip install ecomp

---

CLI Quickstart

All commands are exposed through the ecomp entry point.

# Compress an alignment (produces example.ecomp, optional JSON sidecar)
ecomp zip example.fasta --metadata example.json

# Decompress (writes FASTA by default)
ecomp unzip example.ecomp --alignment-output restored.fasta

# Inspect metadata (summary or JSON)
ecomp inspect example.ecomp --summary

# Diagnostics (Phykit-style aliases in parentheses)
ecomp consensus_sequence example.ecomp             # con_seq
ecomp column_base_counts example.ecomp             # col_counts
ecomp gap_fraction example.ecomp                   # gap_frac
ecomp shannon_entropy example.ecomp                # entropy
ecomp parsimony_informative_sites example.ecomp    # parsimony
ecomp constant_columns example.ecomp               # const_cols
ecomp pairwise_identity example.ecomp              # pid
ecomp alignment_length_excluding_gaps example.ecomp    # len_no_gaps
ecomp alignment_length example.ecomp                   # len_total
ecomp variable_sites example.ecomp                     # var_sites
ecomp percentage_identity example.ecomp                # pct_id
ecomp relative_composition_variability example.ecomp   # rcv

Benchmarks mirror standard codec comparisons:

/usr/bin/time -p ecomp zip data/fixtures/small_phylo.fasta --output out.ecomp
/usr/bin/time -p gzip  -k data/fixtures/small_phylo.fasta
/usr/bin/time -p bzip2 -k data/fixtures/small_phylo.fasta

---

Python API

Everything the CLI does is re-exported in ecomp.

from ecomp import ezip, eunzip, read_alignment, percentage_identity, column_base_counts

# File-based workflow
archive_path, metadata_path = ezip(
    "data/example.fasta",
    metadata_path="data/example.json",  # optional JSON copy
)
restored_path = eunzip(archive_path, output_path="data/restored.fasta")

# Diagnostics on an AlignmentFrame
frame = read_alignment("data/example.fasta")
pct_identity = percentage_identity(frame)
base_counts = column_base_counts(frame)

print(f"Mean pairwise identity: {pct_identity:.2f}%")
print("Column 1 counts:", base_counts[0])

In-memory usage (no intermediate files):

from ecomp import AlignmentFrame, compress_alignment, decompress_alignment

frame = AlignmentFrame(
    ids=["s1", "s2"],
    sequences=["ACGT", "ACGA"],
    alphabet=["A", "C", "G", "T"],
)
compressed = compress_alignment(frame)
restored = decompress_alignment(compressed.payload, compressed.metadata)
assert restored.sequences == frame.sequences

Available functions

Compression & I/O — ezip, eunzip, compress_file, decompress_file, compress_alignment, decompress_alignment, read_alignment, write_alignment, alignment_from_sequences, alignment_checksum

Diagnostics & metrics — column_base_counts, column_gap_fraction, column_shannon_entropy, parsimony_informative_columns, parsimony_informative_site_count, constant_columns, majority_rule_consensus, alignment_length, alignment_length_excluding_gaps, variable_site_count, percentage_identity, relative_composition_variability, pairwise_identity_matrix

Supporting types — AlignmentFrame, CompressedAlignment, PairwiseIdentityResult, __version__

---

Development

make test.fast        # unit + non-slow integration tests
make test             # full test matrix
make lint             # lint checks (ruff, black, isort)
make format           # auto-formatting
mypy ecomp            # optional type checking

Build docs locally:

make docs
open docs/_build/html/index.html

Build and publish distributions:

pip install build twine
python -m build
python -m twine check dist/*
python -m twine upload dist/*

Benchmarking eComp vs. PhyKIT

python scripts/benchmark_metrics.py data/example.ecomp \
    --operations consensus shannon_entropy variable_sites \
    --repeat 5 --warmup 1 --json results.json --csv results.csv

The script runs each metric via the ecomp CLI (on the compressed archive) and the corresponding phykit command on a decompressed alignment, then reports average and best runtimes. Add --json/--csv to emit machine-readable output.

---

License

eComp is released under the MIT License. See LICENSE.