From Intention to Implementation: Automating Biomedical Research via LLMs

This repository is the official implementation of "From Intention to Implementation: Automating Biomedical Research via LLMs".

Framework of BioResearcher

Requirements

To install requirements:

pip install -r requirements.txt

Data Preparation

The input of BioResearcher includes the experimental target, experimental requirements and experimental conditions (optional). For detailed input data format, please refer to bioresearcher/topic/topic-example.json file.

Running BioResearcher

Once your input data is ready, you can run BioResearcher to generate the experimental protocol by executing the following command.

cd bioresearcher

export OPENAI_API_KEY='your_openai_api_key'
export OPENAI_API_KEY='the_openai_base_url'
export NCBI_API_KEY='your_ncbi_api_key'

python main.py 
    --topics_path 'path of inputed topics'
    --paper_ref_path 'data/paper_ref.pdf'
    --paper_ref_fewshots_path 'data/paper_ref_demo.json'
    --search_paper_model 'LLM used for searching papers'
    --search_datasets_model 'LLM used for searching datasets'
    --report_generation_model 'LLM used for generating reports'
    --analyze_model 'LLM used for analyzing reports'
    --design_model 'LLM used for designing protocols'
    --translate_model 'LLM used for translation'
    --dry_experiment_extractor_model 'gpt-4oLLM used for extracting dry experiments'
    --coding_model 'LLM used for generating code'
    --search_paper_retmax 'Maximum number of retrieved papers'
    --search_datasets_retmax 'LLM used for searching datasets'
    --num_protocols 'number of protocols to be generated'