gtars-scoring

Wrapper around gtars-overlaprs for generating X-by-peak count matrices.

Purpose

gtars-scoring wraps the core overlap infrastructure from gtars-overlaprs to produce count matrices for genomic analysis. This module handles all use cases that need X-by-peak matrices:

• Cell-by-peak matrices (single-cell ATAC-seq)
• Sample-by-peak matrices (bulk ATAC-seq/ChIP-seq)
• Pseudobulk-by-peak matrices (aggregated single-cell data)

What it does

Generates count matrices from fragment-peak overlaps where:

• Rows = analysis units (cells, samples, or pseudobulk aggregations)
• Columns = consensus regions (peaks)
• Values = overlap counts

CLI Usage

gtars fscoring <fragments> <consensus> /
  --mode <atac|chip> /
  --output <output.csv.gz>

Arguments:

• fragments: Glob pattern for fragment files (e.g., "*.bed.gz")
• consensus: Path to consensus peak set (BED file)
• --mode: Scoring mode (atac or chip, default: atac)
  • atac: Counts 5' and 3' cut sites with shifts (+4/-5 bp)
  • chip: Counts full fragment overlaps
• --output: Output file path (default: count_matrix.csv.gz)

Example:

gtars fscoring "fragments/*.bed.gz" peaks.bed /
  --mode atac /
  --output counts.csv.gz

Rust Library Usage

use gtars::scoring::{
    region_scoring_from_fragments,
    ConsensusSet,
    FragmentFileGlob,
    ScoringMode
};

// Set up inputs
let mut fragments = FragmentFileGlob::new("path/to/*.bed.gz")?;
let consensus = ConsensusSet::new("peaks.bed".into())?;

// Generate count matrix
let count_matrix = region_scoring_from_fragments(
    &mut fragments,
    &consensus,
    ScoringMode::Atac
)?;

// Write to file
count_matrix.write_to_file("output.csv.gz")?;

// Or access counts directly
let count = count_matrix.get(row, col);

Output Format

Gzipped CSV file with:

• Each row = one fragment file
• Each column = one consensus region
• Values = count of overlapping fragments