GOReasoner

Official code for GOReasoner: Enhancing Protein Function Annotation through Explainable Reasoning with LLM

Requirements

uv pip install -r requirements.txt

Data Preprocessing

python get_testdata.py --input test_proteins.txt --output testdata.json

For test_proteins.txt, each line is [protein id]\t[goid]\t[pmid], for example,

P12345	GO:0042802	12345678
P67890	GO:0003674	11111111

Prediction

prediction and propagation

python GOReasoner.py \
    --input_file testdata/GOR2023/bp_with_desc.json \
    --output_file results/GOR2023/bp_results.json \
    --domain bp

consensus fusion

python consensus_fusion.py \
    --init_res init_res.txt \
    --refine_res refine_res.txt

Models

You can download from huggingface: GOExplainer-Qwen3-8B