pyeris 0.0.1a1

Graph-aware contextual annotation of targeted genomic features

eris 🧬🍏✨

⋆⁺‧₊☽◯☾₊‧⁺⋆ Graph-aware contextual annotation of targeted genomic features. ⋆⁺‧₊☽◯☾₊‧⁺⋆

eris is a bioinformatics pipeline designed to resolve and annotate shattered genomic features (like mobile genetic elements or structural variants) directly from assembly graphs. By employing a dynamic "Anchor-and-Traverse" depth-first search through de Bruijn/string graphs (GFA), eris stitches together split alignments across multiple micro-contigs, seamlessly bridging sequence bubbles and gaps.

Crucially, it utilizes fractional read depth flow to distinguish between dominant wild-type structures and rare sub-clonal insertions, returning detailed evolutionary context including upstream flanks, downstream flanks, and trapped passenger genes.

---

🚀 Features

• Graph-Aware Traversal: Overcomes aligner limitations by traversing assembly graph topology to bridge unaligned micro-contigs and resolve shattered targets.
• Sub-Clonal Resolution: Integrates graph read-depth (dp/rd tags) to calculate fractional copy number, easily distinguishing low-frequency variant bubbles from dominant paths.
• Contextual Annotation: Sweeps across stitched paths to identify internal passenger genes (e.g., AMR genes trapped inside transposons) and flanking genomic context.
• High Performance: Built for speed with a hybrid architecture utilizing numpy, numba, and mappy (Minimap2).
• Standardized Outputs: Generates detailed tabular reports (TSV), locus-specific FASTAs, and standard annotations (pyfgs powered GFF3/FAA).

📦 Installation

eris requires Python 3.11 or later.

Install directly from PyPI:

pip install eris

Dependencies

eris relies on the following core libraries:

• numpy (>=2.4)
• numba (>=0.65)
• mappy (>=2.3)
• pyfgs (>=0.0.1)
• biopython (>=1.87)

🛠️ Usage

eris installs a command-line interface eris for immediate use.

eris -i assembly.gfa -d targets.fasta -o results/sample_A

Basic Arguments

• -i, --genome: Path to a genome in fasta or GFA format (can be compressed). GFA is heavily recommended to utilize topological stitching.
• -d, --targets: Path to target nucleotide features (fasta or pre-indexed .mmi).
• -o, --outprefix: Prefix for all generated output files.
• -f, --feature-type: The type of feature to annotate (default: CDS).
• --hops: Maximum number of contextual genes to sweep upstream/downstream (default: 3).
• --tolerance: Distance tolerance in base pairs for merging clustered alignments (default: 0).

Outputs

Depending on the arguments provided, eris will output:

1. {prefix}_report.tsv: A detailed report of every resolved locus, including targets, context (INSIDE, UPSTREAM, DOWNSTREAM), biological effects, topological hops, and fractional depths.
2. {prefix}_loci.fasta: The stitched nucleotide sequences for each assembled structural variant.
3. {prefix}_assembly.gff: GFF3 annotation of the global features.
4. {prefix}_proteins.faa: Amino acid fasta of the global features.

📚 Documentation

For detailed guides, API reference, and advanced configuration, visit the eris Documentation.

💻 Development

eris uses hatch for build and environment management.

To set up a local development environment (uv needs to be installed):

# Clone the repository
git clone [https://github.com/tomdstanton/eris.git](https://github.com/tomdstanton/eris.git)
cd eris
make dev

Running Tests & Linting

eris enforces rigorous type checking and linting.

# Run tests
pytest

# Run static type checking
mypy

# Run linting and formatting
ruff check
ruff format

📝 License

This project is licensed under the terms of the LICENSE file included in the repository.

🤝 Authors

• Tom Stanton - [e-mail]

---