Skip to content

Latest commit

 

History

History
64 lines (46 loc) · 2.7 KB

readme.md

File metadata and controls

64 lines (46 loc) · 2.7 KB

GEO Metadata Extractor

This project is designed to extract and analyze metadata from Gene Expression Omnibus (GEO) studies, focusing on human (Homo sapiens) samples. It uses a combination of database operations, natural language processing, and machine learning techniques to process and interpret the metadata.

Key Components

  1. Database Management

    • Uses DuckDB to store and query GEO metadata
    • Creates and manages tables for both raw metadata and parsed results

  2. Metadata Extraction

    • Extracts relevant information from XML files
    • Focuses on human studies by filtering for "Homo sapiens" organisms

  3. Text Processing

    • Generates text descriptions for each study based on various metadata fields
    • Includes information such as title, summary, overall design, treatment protocols, and more

  4. LLM-based Information Extraction

    • Utilizes Language Models (LLMs) for advanced information extraction
    • Supports multiple LLM providers including Groq and Azure OpenAI
    • Extracts specific fields like high-level indication, drug exposure, modalities, etc.

  5. Langfuse Integration

    • Incorporates Langfuse for prompt management and observability

  6. Extensible Architecture

    • The Extractor class serves as a base for creating specialized extractors
    • GSEmetaExtractor is an example of a specialized extractor for GEO metadata

Key Features

  • Processes GEO studies in batches
  • Saves intermediate results to avoid redundant processing
  • Provides progress tracking and error handling
  • Allows for easy extension to extract different types of information

Usage

  1. Ensure the necessary environment variables are set (e.g., API keys for LLM providers)
  2. Run the metadata extraction script to populate the DuckDB database
  3. Use the GSEmetaExtractor or create custom extractors to process the metadata
  4. Analyze the extracted information stored in the database and JSON files

Requirements

  • Python 3.x
  • DuckDB
  • Langfuse
  • OpenAI or Groq API access

Environment Setup

This project uses environment variables for configuration. Create a .env file in the root directory of the project and add the following variables:

  • GROQ_API_KEY: API key for Groq
  • LANGFUSE_SECRET_KEY: Secret key for Langfuse
  • LANGFUSE_PUBLIC_KEY: Public key for Langfuse
  • LANGFUSE_HOST: Host URL for Langfuse
  • AZURE_OPENAI_ENDPOINT: Endpoint URL for Azure OpenAI
  • AZURE_OPENAI_API_KEY: API key for Azure OpenAI
  • OPENAI_API_VERSION: Version of the OpenAI API to use

Ensure all necessary API keys and endpoints are properly set in the .env file before running the project. The actual values for these variables should be kept confidential and never committed to version control.