gpredomicspy

Python bindings for the gpredomics Rust engine.

gpredomics discovers sparse, interpretable predictive models using BTR (Binary/Ternary/Ratio) languages. Designed for omics/metagenomics data but applicable to any binary classification task.

Requirements

• Python >= 3.8
• Rust toolchain (for building from source)
• numpy, pandas

Installation

# From source (requires Rust toolchain)
pip install .

# Or using maturin for development
pip install maturin
maturin develop

Quick Start

from gpredomicspy import Param, fit

# Load parameters
param = Param()
param.load("param.yaml")
param.set("max_epochs", 50)
param.set("population_size", 5000)

# Run
experiment = fit(param)

# Results
best = experiment.best_population().best()
metrics = best.get_metrics()
print(f"AUC: {metrics['auc']:.4f}")
print(f"Features: {best.get_features()}")

Development Status

This package is in early development (Phase 1: Core Bindings).

License

GPL-3.0 - see gpredomics for details.