AI Gene Enrichment

A comprehensive Python package for performing gene enrichment analysis using multiple bioinformatics tools and AI-powered synthesis. This tool integrates Enrichr, ToppFun, gProfiler, literature search, and AI summarization to provide detailed biological insights from gene lists.

License

This project is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC-BY-NC 4.0). This means you are free to:

• Share — copy and redistribute the material in any medium or format
• Adapt — remix, transform, and build upon the material

Under the following terms:

• Attribution — You must give appropriate credit and indicate if changes were made
• NonCommercial — You may not use the material for commercial purposes

For commercial licensing inquiries, please contact: apetti@mgh.harvard.edu

Installation

From PyPI (Recommended)

pip install ai-gene-enrichment

From Source

1. Clone this repository:

   git clone https://github.com/calvinrboyce/ai-gene-enrichment.git
   cd ai-gene-enrichment
   

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install in development mode:

   pip install -e .
   

API Key Configuration

Regardless of how you install the package, you will need to set up your API keys in your project directory:

# Create a .env file
echo "OPENAI_API_KEY=your_openai_api_key_here" >> .env
echo "ENTREZ_API_KEY=your_entrez_api_key_here" >> .env

Usage

AI synthesis requires an API key to access OpenAI's models. API keys can be obtained by PIs or other researchers by setting up an organization at platform.openai.com and adding an API key at platform.openai.com/settings/organization/api-keys. Note that using more sophisticated models beyond the default may require organization verification and you may have to earn a higher usage tier with OpenAI. With the default model you can expect to spend <$0.03 per gene list

The model also requires an API key from NCBI for literature results which can be obtained by creating an account with NCBI here

Basic Usage

import os
import dotenv
from aige import AIGeneEnrichment

# Load environment variables
dotenv.load_dotenv()
openai_api_key = os.getenv("OPENAI_API_KEY")
entrez_api_key = os.getenv("ENTREZ_API_KEY")

# Initialize the agent
aige = AIGeneEnrichment(
    open_ai_api_key=openai_api_key,
    entrez_api_key=entrez_api_key
)

# Define your gene list
genes = ["ANLN", "CENPF", "NUSAP1", "TOP2A", "CCNB1", "PRC1", "TPX2", "UBE2C", "BIRC5"]

# Run analysis
results = aige.run_analysis(
    genes=genes,
    email="your.email@example.com",  # Required for NCBI literature search
    background_genes=[],  # Optional background gene set for enrichment analysis
    ranked=False,  # Set to True if genes are ranked by differential expression
    search_terms=["cancer", "metastasis", "cell cycle"],  # Optional literature search terms
    context="This gene list characterizes a cluster of cells in a brain metastasis"
)

# Access results
print(results['summary'])
for theme in results['themes']:
    print(f"Theme: {theme['theme']}")
    print(f"Description: {theme['description']}")
    print()

Advanced Configuration

# Customize enrichment sources and parameters
aige = AIGeneEnrichment(
    open_ai_api_key=openai_api_key,
    entrez_api_key=entrez_api_key,
    open_ai_model="gpt-5-mini",
    results_dir="custom_results",
    enrichr_sources={
        "CellMarker_2024": "CellMarker",
        "MSigDB_Hallmark_2020": "MSigDB-H",
        "ChEA_2022": "ChEA"
    },
    gprofiler_sources=['HPA', 'MIRNA'],
    toppfun_sources=['TFBS', 'MicroRNA'],
    terms_per_source=20,  # Number of terms to retrieve per source
    papers_per_gene=3,    # Number of papers to retrieve per gene
    aggregate_papers=15   # Number of papers to retrieve for full gene list
)

# Example with background gene set
background_genes = ["GENE1", "GENE2", "GENE3", ...]  # Your background gene set
results = agent.run_analysis(
    genes=genes,
    email="your.email@example.com",
    background_genes=background_genes,  # Use custom background for enrichment
    ranked=True,
    search_terms=["cancer", "metastasis"],
    context="Analysis with custom background gene set",
    save_results=2,     # Save excel file of results and JSON files of all intermediate results
    verbose=False,      # Don't print status updates
    analysis_name="Metastasis Enrichment"
)

Background Gene Sets

The agent supports custom background gene sets for enrichment analysis, which can improve the statistical significance and biological relevance of your results. If a background gene set is available, please use it.

Note: When no background genes are provided (empty list), the tools use their default reference sets.

Working with Results

The analysis returns a structured dictionary with the following components:

results = {
    'themes': [
        {
            'theme': 'Cell Cycle Regulation',
            'description': 'Genes involved in cell cycle control and progression...',
            'terms': ['GO:0007049', 'KEGG:04110', ...]
        }
    ],
    'summary': 'Comprehensive analysis summary...'
}

Output Files

The save_results parameter in run_analysis controls what files are saved:

• save_results=0: No files are saved.
• save_results=1: Saves an Excel spreadsheet (*_enrichment_analysis.xlsx) with the themes and run metadata.
• save_results=2 (default): Saves the Excel spreadsheet AND intermediate JSON files for metadata, Enrichr, ToppFun, gProfiler, literature results, and gene summaries.

Dependencies

• openai>=1.0.0: OpenAI API integration
• requests>=2.31.0: HTTP requests
• python-dotenv>=1.0.0: Environment variable management
• biopython>=1.82: PubMed literature search
• gprofiler-official>=1.0.0: gProfiler API client
• openpyxl>=3.1.5: Excel file generation

API Reference

AIGeneEnrichment

Main class for running gene enrichment analysis workflows.

Constructor Parameters

• open_ai_api_key (str): OpenAI API key (required)
• entrez_api_key (str): NCBI Entrez API key for literature search (required)
• open_ai_model (str): OpenAI model to use (default: "gpt-5-mini")
• results_dir (str): Directory to save results (default: "aige_results")
• enrichr_sources (dict): Enrichr sources to use. Dictionary should map the name of the source (found in Enrichr's documentation) to an abbreviated version to be used in the analysis (default: {})
• gprofiler_sources (list): gProfiler sources to use. List should include terms from ['TF', 'MIRNA', 'HPA', 'CORUM'] to be used in the analysis (default: [])
• toppfun_sources (list): ToppFun sources to use. List should include terms from ['MousePheno', 'Domain', 'Pubmed', 'Cytoband', 'TFBS', 'GeneFamily', 'Coexpression', 'CoexpressionAtlas', 'ToppCell', 'Computational', 'MicroRNA', 'Drug', 'Disease'] to be used in the analysis (default: ['ToppCell'])
• terms_per_source (int): Number of terms per source (default: 20)
• papers_per_gene (int): Number of papers to fetch for each gene (default: 0). For large gene lists this is not recommended as it may overwhelm the context window of the LLM
• aggregate_papers (int): Number of papers to fetch that mention any subset of all genes provided (default: 15)

Methods

run_analysis(genes, email, background_genes=[], ranked=True, search_terms=[], context="None", save_results=True, analysis_name=None)

Run the complete gene enrichment analysis workflow.

Parameters:

• genes (List[str]): List of gene symbols to analyze (required)
• email (str): Email address for NCBI's reference (required for literature search)
• background_genes (List[str]): List of background genes to use for enrichment analysis (default: [])
• ranked (bool): Whether the genes are ranked by differential expression (default: False)
• search_terms (List[str]): List of search terms to use in literature search (default: [])
• context (str): Context of where the genes came from and what you're studying (default: "None")
• save_results (int): Level of results saving - 0 for none, 1 for Excel, 2 for Excel + intermediate JSON files (default: 2)
• verbose (bool): Whether to print status updates during enrichment
• analysis_name (str): Name for the analysis run, used in file naming (default: None)

Returns:

• Dict[str, Any]: Themed results dictionary containing:
  • themes: List of identified biological themes, each with:
    • theme: Theme name
    • description: Description of the theme's function and identification rationale
    • confidence: Confidence score (0-1)
    • terms: List of enrichment terms that contributed to theme identification
  • summary: Comprehensive analysis summary
  • metadata: Dictionary of enrichment metadata