pyrrblup 1.0.0

Python implementation of core rrBLUP workflows with R-aligned baselines

pyrrBLUP

Python implementation of core rrBLUP workflows with numeric checks against the R package.

Installation

pip install pyrrblup

Quick Start

import numpy as np
import pandas as pd

from pyrrblup import amat, kin_blup, mixed_solve

markers = pd.DataFrame(
    {
        "m1": [0, 0, 2, 2],
        "m2": [0, 2, 0, 2],
        "m3": [0, 1, 1, 2],
    },
    index=["s1", "s2", "s3", "s4"],
)

kinship = amat(markers).matrix
y = np.array([1.0, 2.0, 1.5, 2.5])
X = pd.DataFrame(
    {
        "Intercept": [1.0, 1.0, 1.0, 1.0],
        "covariate": [0.0, 1.0, 0.5, 1.5],
    }
)
Z = pd.DataFrame(
    {
        "line_A": [1.0, 0.0, 1.0, 0.0],
        "line_B": [0.0, 1.0, 0.0, 1.0],
    }
)
K = np.array(
    [
        [1.0, 0.2],
        [0.2, 1.0],
    ]
)

print(kinship.shape)
fit = mixed_solve(y, X=X, Z=Z, K=K)
print(fit.beta_terms, fit.beta)
print(fit.Vu, fit.Ve)

Run kin_blup from file paths:

from pyrrblup import kin_blup

result = kin_blup("genotype.csv", "phenotype.csv", "loc_BeiJ")
print(result.predictions.head())

mixed_solve follows an rrBLUP-like matrix interface:

result = mixed_solve(y, Z=None, K=None, X=None)

kin_blup accepts either:

• genotype and phenotype file paths
• genotype and phenotype pandas.DataFrame objects

Both inputs must be the same kind. Mixed input types are rejected.

Real Data Examples

This repository also includes deterministic subsets cut from the real loc_BeiJ data.

• examples/minimal_real_data/

  • 32 samples
  • 64 markers
  • intended for quick README-scale runs

• examples/medium_real_data/

  • 256 samples
  • 512 markers
  • intended for more realistic local demonstrations

Build or refresh the example data:

python scripts/build_examples.py

Run the minimal real-data example:

python examples/minimal_real_data/run_example.py

More detail:

• docs/mixed_solve.md
• examples/README.md

Development

Install from source:

pip install -e .

Run the test suite with:

OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 MKL_NUM_THREADS=1 python -m pytest -v