Command-line Usage
==================
.. code-block:: bash

    python -m articull classify \
            <input_file> <output_prefix> <model_dir> <bams> \
            [--resources_dir <path>] [--chunksize <n>] [--features_file <path>] [--cores <ncores>]

Arguments
^^^^^^^^^
- **<input_file>** (required): MAF or VCF file containing candidate variants
- **<output_prefix>** (required): Output prefix (directory and sample name, e.g., /path/to/sample1. Output files will be saved as /path/to/sample1_features.tsv and /path/to/sample1_result.tsv)
- **<model_dir>** (required): Directory containing `model.pkl` and `scaler.pkl`
- **<bams>** (required): List of BAM files containing sequencing data
- **-\-resources_dir** (optional): Path to directory containing folder of mappability tracks (default: `resources/hg19_mappability.bedGraph`)
- **-\-chunksize** (optional): Number of rows per worker for parallel processing (default: 5000)
- **-\-features_file** (optional): File containing features (e.g., generated by a previous run of ArtiCull). If not provided, features will be extracted from input BAM files
- **-\-cores** (optional): Number of CPU cores for parallel processing (default: all available cores)
- **-\-no_vcf_filter** (optional): Do not filter variants in VCF input file (default: filter out variants with FILTER field not equal to "PASS")
- **-\-extract_features_only** (optional): Extract features but do not classify variants.