HARP

HARP — Haplotype Allele Read Profiler.
A Python tool for counting haplotype-specific read support for SNVs from a haplotagged BAM file and a phased VCF file.

---

Features

• Computes allele counts per haplotype (h1 and h2) for SNVs.
• Supports parallel processing of BAM files for faster execution.
• Handles indexed BAM and bgzipped VCF files.
• Produces a simple TSV output suitable for downstream analysis.

---

Installation

Using Poetry:

git clone https://github.com/yourusername/harp.git
cd harp
poetry install

Usage

Run HARP via CLI:

poetry run harp run --bam PATH_TO_BAM --vcf PATH_TO_VCF --out OUTPUT_TSV [--threads N]

Example

poetry run harp run \
    --bam specification/test_data/giab_2023.05.hg002.haplotagged.chr16_28000000_29000000.processed.30x.bam \
    --vcf specification/test_data/giab_2023.05.hg002.wf_snp.chr16_28000000_29000000.vcf.gz \
    --out result.tsv \
    --threads 4

CLI options:

• bam FILE — Input BAM file (indexed) [required]
• vcf FILE — Input phased VCF file [required]
• out FILE — Output TSV file [required]
• threads INTEGER — Number of parallel threads to use (default: 4)

Output Format

The output is a tab-separated file with the following columns:

• chrom — Chromosome name pos — 0-based genomic position of the variant
• h1_REF — Number of reads supporting the reference allele on haplotype 1
• h1_ALT — Number of reads supporting the alternate allele on haplotype 1
• h2_REF — Number of reads supporting the reference allele on haplotype 2
• h2_ALT — Number of reads supporting the alternate allele on haplotype 2

Example output:

chrom	pos	    h1_REF	h1_ALT	h2_REF	h2_ALT
chr16	28001380	11	    4	    3	    12
chr16	28002343	11	    5	    3	    12
chr16	28003017	12	    4	    3	    13
chr16	28003087	12	    4	    3	    11
chr16	28004800	13	    4	    3	    9
...

Notes

• HARP assumes that the BAM file is haplotagged (HP tag present).
• The BAM file must be coordinate-sorted and indexed (.bai present). Unsorted BAMs will lead to incorrect counts or skipped reads.
• Only bi-allelic SNVs are counted.
• Chunk size and threading can be tuned for performance.