---
name: gtars
description: A high-performance Rust toolkit (with Python bindings and a CLI) for genomic interval analysis; use it when you need fast overlap queries, coverage track generation, genomic tokenization for ML, reference sequence verification, or fragment processing.
license: MIT
author: AIPOCH
---
> **Source**: [https://github.com/aipoch/medical-research-skills](https://github.com/aipoch/medical-research-skills)

## When to Use

- **Overlap and set operations on genomic intervals** (e.g., peak/promoter overlap, variant annotation, shared-feature detection).
- **Coverage track generation** from interval-like inputs (e.g., ATAC-seq/ChIP-seq/RNA-seq coverage for visualization in genome browsers).
- **Machine-learning preprocessing** where genomic regions must be converted into discrete **tokens** (e.g., Transformer-style models, geniml-style pipelines).
- **Reference sequence management and verification** (e.g., subsequence retrieval, digest calculation aligned with GA4GH refget concepts).
- **Single-cell fragment workflows** (e.g., splitting fragments by barcode/cluster, scoring fragments against reference region sets).

## Key Features

- **Rust performance** with low overhead; designed for large genomic datasets.
- **Python bindings** for integration into analysis notebooks/pipelines.
- **CLI tooling** for batch processing and shell workflows.
- **Fast overlap detection** via IGD-style indexing and interval operations.
- **Coverage track generation** (WIG/BigWig workflows via the `uniwig` functionality).
- **Genomic tokenizers** for ML-ready representations of genomic regions.
- **Reference sequence utilities** (FASTA-backed stores, subsequence retrieval, digesting).
- **Fragment processing and scoring** for common single-cell genomics tasks.

> Additional module-specific guidance may be available in: `references/overlap.md`, `references/coverage.md`, `references/tokenizers.md`, `references/refget.md`, `references/python-api.md`, and `references/cli.md`.

## Dependencies

- **Python package**: `gtars` (version not specified in the source document)
- **Rust toolchain** (for CLI install): `cargo` (version not specified)
- **Rust crate**: `gtars = "0.1"` (as shown in the example)

## Example Usage

### Python: overlap analysis workflow (runnable)

```python
import gtars

# Load two region sets
peaks = gtars.RegionSet.from_bed("chip_peaks.bed")
promoters = gtars.RegionSet.from_bed("promoters.bed")

# Find overlaps (peaks that overlap promoters)
overlapping_peaks = peaks.filter_overlapping(promoters)

# Export results
overlapping_peaks.to_bed("peaks_in_promoters.bed")
```

### CLI: generate coverage tracks (runnable)

```bash
# Generate WIG coverage at a given resolution
gtars uniwig generate --input atac_fragments.bed --output coverage.wig --resolution 10

# Generate BigWig coverage for genome browser visualization
gtars uniwig generate --input atac_fragments.bed --output coverage.bw --format bigwig
```

### Python: ML tokenization (runnable)

```python
import gtars
from gtars.tokenizers import TreeTokenizer

# Load regions and build a tokenizer from BED
regions = gtars.RegionSet.from_bed("training_peaks.bed")
tokenizer = TreeTokenizer.from_bed_file("training_peaks.bed")

# Tokenize each region into a discrete representation
tokens = [tokenizer.tokenize(r.chromosome, r.start, r.end) for r in regions]

print(tokens[:5])
```

## Implementation Details

- **Interval overlap & indexing**: Overlap queries are designed around an IGD-like index to accelerate repeated interval queries (build once, query many). Typical parameters are chromosome, start, end; results are overlapping intervals or derived set operations.
- **Coverage generation (`uniwig`)**: Produces coverage tracks from interval/fragments input. Common knobs include output format (e.g., WIG vs BigWig) and resolution/binning for track granularity.
- **Tokenization**: Tokenizers (e.g., `TreeTokenizer`) map genomic coordinates to discrete tokens suitable for ML pipelines. Token vocabularies are commonly derived from a BED-defined training region universe.
- **Reference sequence store**: FASTA-backed reference access supports subsequence retrieval and digesting/verification workflows aligned with refget-style usage.
- **Fragment workflows**: Fragment splitting and scoring operate on fragment-like inputs (often TSV/BED-style) and can be used for barcode/cluster partitioning and enrichment-style scoring against reference region sets.

---
Source attribution: aipoch
Original project: https://github.com/aipoch/medical-research-skills/blob/686e09dfdb45d9fa1a08664f06bb93769b3c9d91/scientific-skills/Data%20Analysis/gtars/SKILL.md
License: MIT
License information: https://opensource.org/license/mit
Original upstream instructions; downloaded from Mega AI Hub.

## Upstream license notice

MIT License

Copyright (c) 2026 AIpoch

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.



# Mega AI Hub resource metadata

Creator / publisher: aipoch
Original project: https://github.com/aipoch/medical-research-skills/blob/686e09dfdb45d9fa1a08664f06bb93769b3c9d91/scientific-skills/Data%20Analysis/gtars/SKILL.md
Attribution basis: Source repository record

Fields of work: Data & Analytics, Development & Engineering, Education & Research, Product Management
Roles: Business Intelligence Analyst, Data Scientist, ML / AI Engineer, Product Manager, Researcher
Specialties: API, Backend
Tools: Python, Rust
Country applicability: Country not specified
Classification basis: Source catalog labels
Instruction SHA-256: 691f4aa52689c8cd42851d9b58b3df184b17477460ec04e0963e2297bba2639c