AMP-SEMiner: AntiMicrobial Peptide Structural Evolution Miner

This repository contains the custom code for AMP-SEMiner, a robust and comprehensive AI framework designed for residue-level AMP discovery. The framework accurately identifies specific residues within a protein as either AMP or non-AMP.

The AMPs identified using AMP-SEMiner are available through our dedicated data portal, MAG-AMPome (http://mag-ampome.aigene.org.cn). Visit the portal to access this comprehensive resource of antimicrobial peptides.

AMP-SEMiner is run on Python 3.10 and PyTorch 2.1.2. You can build a conda environment for AMP-SEMiner using this script.

The model weights of AMP-SEMiner and minimum datasets to run AMP-SEMiner are available in Zenodo (DOI: 10.5281/zenodo.14348290). A small example dataset are also provided:

Tok_CLS.tar.gz

Tok_CLS_LoRA.tar.gz

2_steps.tar.gz

example_dataset.tar.gz

mkdir model_weights
cd model_weights

## download model weights
wget https://zenodo.org/records/14348290/files/Tok_CLS.tar.gz
tar -zxvf Tok_CLS.tar.gz

wget https://zenodo.org/records/14348290/files/Tok_CLS_LoRA.tar.gz
tar -zxvf Tok_CLS_LoRA.tar.gz

cd -

## download example datasets
wget https://zenodo.org/records/14348290/files/example_dataset.tar.gz
tar -zxvf example_dataset.tar.gz

mkdir example_out

#conda activate ampseminer
python End_to_end_Tok_CLS/pred_Token.py \
    --input example_dataset/test_dataset.csv \
    --output example_out/test.out_pred.tsv \
    --model_name model_weights/Tok_CLS/epoch15 \
    --batch_size 4 \
    --max_len 300

Example datasets for AMP model trainining:

example_data.tar.gz

example_labels.tar.gz

## download example datasets for AMP model training

wget https://zenodo.org/records/14348290/files/example_data.tar.gz
tar -zxvf example_data.tar.gz

wget https://zenodo.org/records/14348290/files/example_labels.tar.gz
tar -zxvf example_labels.tar.gz

#conda activate ampseminer
python End_to_end_Tok_CLS/train_Tokens.py \
    --data_path ./example_data \
    --label_path ./example_labels \
    --model_name facebook/esm2_t30_150M_UR50D \
    --num_classes 2 \
    --ft_mode full \
    --epochs 10 \
    --batch_size 16 \

Install and run our AMP discovery model run_pred_AMP_SEMiner.ipynb

zhengjinfang1220@gmail.com, liwh@zhejianglab.org, liwh@tongji.edu.cn