htsgetr

A Rust implementation of the htsget protocol (v1.3) for serving genomic data over HTTP. Built with noodles for bioinformatics I/O and axum for the web server.

Features

• htsget 1.3 compliant - GET/POST endpoints for reads and variants
• Multiple formats - BAM, CRAM, VCF, BCF via noodles
• Extensions - FASTA/FASTQ support beyond the spec
• Multiple storage backends - Local filesystem, S3, and HTTP/HTTPS
• JWT authentication - Optional Bearer token auth with JWKS/static keys
• Python bindings - PyO3 integration via maturin
• Async - Built on tokio for high concurrency

Installation

From source (Rust)

git clone https://github.com/bolu-atx/htsgetr.git
cd htsgetr
cargo install --path .

Python

pip install maturin
maturin develop --features python

Quick Start

# Create a data directory with some BAM/VCF files
mkdir -p data
cp /path/to/sample.bam data/
cp /path/to/sample.bam.bai data/

# Start the server
htsgetr --data-dir ./data

# Query reads
curl http://localhost:8080/reads/sample

Usage

Server

# Start server with default settings
htsgetr --data-dir /path/to/data

# Custom host/port
htsgetr --host 0.0.0.0 --port 8080 --data-dir /path/to/data

# With explicit base URL (for proxied setups)
htsgetr --data-dir /path/to/data --base-url https://example.com/htsget

Configuration

S3 Storage

cargo build --features s3

HTSGET_STORAGE=s3 \
HTSGET_S3_BUCKET=my-genomics-bucket \
HTSGET_S3_PREFIX=samples/ \
htsgetr

HTTP Storage

cargo build --features http

HTSGET_STORAGE=http \
HTSGET_HTTP_BASE_URL=https://files.example.com/genomics/ \
htsgetr

Authentication

Enable JWT/Bearer token authentication by building with the auth feature:

cargo build --features auth

HTSGET_AUTH_ENABLED=true \
HTSGET_AUTH_ISSUER=https://auth.example.com \
htsgetr --features auth

When auth is enabled:

• Public endpoints (root, service-info) don't require authentication
• All other endpoints require a valid Authorization: Bearer <token> header
• Data URLs in tickets are HMAC-signed with expiry to prevent unauthorized access

Data Directory Structure

Place files in the data directory with standard extensions:

data/
├── sample1.bam
├── sample1.bam.bai
├── sample2.vcf.gz
├── sample2.vcf.gz.tbi
├── reference.fa
└── reference.fa.fai

API Reference

Reads Endpoint

# Get all reads
curl http://localhost:8080/reads/sample1

# Get reads for a region
curl "http://localhost:8080/reads/sample1?referenceName=chr1&start=0&end=1000000"

# Header only
curl "http://localhost:8080/reads/sample1?class=header"

# Request CRAM format
curl "http://localhost:8080/reads/sample1?format=CRAM"

# POST with multiple regions
curl -X POST http://localhost:8080/reads/sample1 \
  -H "Content-Type: application/json" \
  -d '{
    "format": "BAM",
    "regions": [
      {"referenceName": "chr1", "start": 0, "end": 1000000},
      {"referenceName": "chr2", "start": 0, "end": 500000}
    ]
  }'

Variants Endpoint

# Get all variants
curl http://localhost:8080/variants/sample2

# Get variants for a region
curl "http://localhost:8080/variants/sample2?referenceName=chr1&start=0&end=1000000"

Sequences Endpoint (Extension)

# Get FASTA sequence
curl http://localhost:8080/sequences/reference

Service Info

curl http://localhost:8080/service-info

Response Format

Successful responses return a JSON ticket per the htsget spec:

{
  "htsget": {
    "format": "BAM",
    "urls": [
      {
        "url": "http://localhost:8080/data/reads/sample1",
        "class": "body"
      }
    ]
  }
}

Error responses:

{
  "htsget": {
    "error": "NotFound",
    "message": "not found: sample1"
  }
}

Python Bindings

from htsgetr import HtsgetServer, HtsgetClient

# Start a server
server = HtsgetServer("/path/to/data", port=8080)
server.run()  # Blocking

# Use the client
client = HtsgetClient("http://localhost:8080")
ticket = client.reads("sample1", reference_name="chr1", start=0, end=1000000)

Roadmap

• Server scaffold with axum
• htsget 1.3 types and error handling
• Local filesystem storage backend
• Reads endpoint (BAM/CRAM)
• Variants endpoint (VCF/BCF)
• Sequences endpoint (FASTA/FASTQ extension)
• Index-based byte range queries (BAI, TBI, CSI)
• S3 storage backend with pre-signed URLs
• HTTP/HTTPS storage backend
• JWT/Bearer token authentication
• CRAM reference resolution
• GCS storage backend
• Python bindings (full implementation)
• Docker image

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Related Projects

• htsget-rs - Another Rust htsget implementation
• noodles - Bioinformatics I/O libraries in Rust
• htslib - C library for HTS formats

License

MIT License - see LICENSE for details.