BioPromptX

Large Language Models for Efficient Biological Information Extraction

No file chosen

About the project

Introduction

BioPromptX provides a method for extracting gene functional descriptions, enzyme kinetic parameters, and related information from scientific articles. Users can also define custom extraction tasks. BioPromptX generates high-quality prompts using a reinforcement learning approach, which are then used to extract the required information effectively.


How to use it

We currently support users in uploading PDF articles online, selecting an LLM, choosing a predefined extraction task or defining a custom one, selecting the output file format, and providing an email address. Once all uploaded articles have been processed, the results will be delivered to the user via email.


How it was made

The method leverages state-of-the-art open-source LLMs for information extraction, including Llama 3.1-70B and DeepSeek-V3. Our vision for BioPromptX is to partially replace manual reading, thereby accelerating data extraction, database construction, and error correction.


Citation

To cite this resource, please use: