omix 0.1.0

A modular Python package for fetching, enriching, and analyzing omics metadata and publications.

omix

A Python package that:

• Fetches comprehensive metadata from public databases (ENA, and soon others).
• Enriches coordinates, dates, host/environment categories, and experimental protocols.
• Searches across multiple publication sources (Crossref, Europe PMC, NCBI, Semantic Scholar, etc.).
• Extracts methodology from full‑text using LLMs (optional).
• Validates findings against reference databases (e.g., primer databases for 16S).
• Works for any omics via plugins.

Installation

pip install omix
# with LLM support:
pip install omix[llm]

Quick Start

Command Line

# Enrich a metadata file with ENA data
omix fetch-metadata samples.tsv --email [e-mail]

# Fetch publications for one or more accessions
omix fetch-publications PRJNA864623 --omics 16S --api-key $LLM_KEY

# Run the full metadata cleaning and enrichment pipeline
omix run-pipeline metadata.csv -o enriched.csv

# NEW: Unified pipeline (metadata + publications + validation + integration)
omix enrich-with-publications samples.csv -o enriched_complete.csv --config config.yaml

Unified Metadata + Publications Pipeline

The enrich-with-publications command provides an end-to-end workflow:

1. Metadata Enrichment: Fetches comprehensive data from ENA (sequences, samples, runs)
2. Publication Discovery: Searches across 12+ publication APIs (Crossref, EuropePMC, NCBI, Semantic Scholar, arXiv, bioRxiv, CORE, DataCite, DOAJ, PLOS, Unpaywall, Zenodo)
3. Publication Validation: Filters to only include publications with direct accession mentions
4. Integration: Merges publication counts and DOIs into the enriched metadata

Output includes all ENA metadata fields plus:

• publication_count: Number of validated publications per study
• publication_dois: Semicolon-separated list of publication DOIs

# Basic usage
omix enrich-with-publications input.csv -o output.csv

# With debug config for faster testing
omix enrich-with-publications input.csv -o output.csv --config config.debug.yaml

# Skip validation (keep all publications found)
omix enrich-with-publications input.csv -o output.csv --no-validate

# With LLM-based methodology extraction
omix enrich-with-publications input.csv -o output.csv --api-key $LLM_KEY

Python API

from omix import Config
from omix.metadata.file_workflow import enrich_metadata_from_path
import asyncio

config = Config(email="[e-mail]")
df = asyncio.run(enrich_metadata_from_path("samples.csv", config=config))
print(df.head())

Configuration

omix can be configured via a YAML file:

credentials:
  email: "[e-mail]"
  ena_email: "[e-mail]"
  llm_api_key: "sk-..."
  ncbi_api_key: "..."

apis:
  sequence:
    ena:
      enabled: true
      max_concurrent: 5
      batch_size: 100
      cache_ttl_days: 30
      fetch_phases: true

metadata:
  sample_id_column: "#sampleid"
  exclude_host: false

paths:
  cache_dir: ".cache"
  logs_dir: "logs"
  primer_db: null

Pass it with --config my_config.yaml or set environment variables like OMIX_EMAIL.