Gtars: Genomic Tools and Algorithms in Rust

Overview

Gtars (from databio, the lab behind geniml) is a high-performance Rust toolkit for manipulating, analyzing, and processing genomic interval data. Its primary purpose is to be the performance-critical backend for geniml, a Python library for machine learning on genomic intervals. It provides overlap/set operations, IGD overlap indexing, coverage (uniwig) tracks, region tokenization for ML, single-cell fragment pseudobulking, and GA4GH refget sequence-collection management.

Use this skill when working with:

• Genomic interval files (BED) — overlaps, jaccard, set ops, coverage
• IGD indexing for fast overlap queries over large interval databases
• Coverage / accumulation tracks via uniwig
• Genomic ML preprocessing and region tokenization
• Single-cell fragment files (split into pseudobulks by cluster)
• Reference sequence digests and retrieval (refget)

Version note: examples are verified against the gtars Python package v0.8 (PyPI). The Python API is exposed through submodules — gtars.models, gtars.tokenizers, gtars.refget, gtars.utils — NOT as flat top-level functions. There is no gtars.igd or gtars.uniwig Python submodule; IGD building and uniwig track generation are CLI-only.

Installation

Python package