DMLF: Enzyme Commission Number Predicting and Benchmarking with Multi-agent Dual-core Learning

This repo contains source codes for a EC prediction tool namely ECRECer, which is an implementation of our paper: 「Enzyme Commission Number Prediction and Benchmarking with Hierarchical Dual-core Multitask Learning Framework」

Detailed information about the framework can be found in our paper

1. Zhenkun Shi, Qianqian Yuan, Ruoyu Wang, Hoaran Li, Xiaoping Liao*, Hongwu Ma* (2022). ECRECer: Enzyme Commission Number Recommendation and Benchmarking based on Multiagent Dual-core Learning. arXiv preprint arXiv:2202.03632.

2. Zhenkun Shi, Rui Deng, Qianqian Yuan, Zhitao Mao, Ruoyu Wang, Haoran Li, Xiaoping Liao*, Hongwu Ma* (2023). Enzyme Commission Number Prediction and Benchmarking with Hierarchical Dual-core Multitask Learning Framework. Research.

Usage

For simply use our tools to predict EC numbers, please visit ECRECEer websiet at https://ecrecer.biodesign.ac.cn

For users who want to run ECRECer locally, please follow the steps below:

We provide docker image and singularity image for users to run ECRECer locally.

Docker image:

# 1. pull ecrecer docker image
docker pull kingstdio/ecrecer

# 2. run ecrecer docker image
# gpu version:
sudo docker run -it -d --gpus all  --name ecrecer -v ~/:/home/ kingstdio/ecrecer #~/ is your fasta file folder
# cpu version:
sudo docker run -it -d --name ecrecer -v ~/:/home/ kingstdio/ecrecer  #~/ is your fasta file folder

# 3. run ECRECer prediction 

sudo docker exec ecrecer python /ecrecer/production.py -i /home/input_fasta_file.fasta -o /home/output_tsv_file.tsv -mode h -topk 10

#-topk: top k predicted EC numbers
#-mode p: prediction mode, predict EC numbers only
#-mode r: recommendation mode, recommend EC numbers with predicted probabilities, the higher the better
#-mode h: hybird mode, use prediction, recommendation and sequence alignment methods

Singularity image:

# 1. pull ecrecer singularity image

# Image ~= 11GB, may take a while to download
wget -c https://tibd-public-datasets.s3.us-east-1.amazonaws.com/ecrecer/sifimages/ecrecer.sif

# 2. run ecrecer singularity image
# gpu version:
singularity run --nv ecrecer.sif python /ecrecer/production.py -i input_fasta_file.fasta -o output_tsv_file.tsv -mode h -topk 10
# cpu version:
singularity run ecrecer.sif python /ecrecer/production.py -i input_fasta_file.fasta -o output_tsv_file.tsv -mode h -topk 10

#-topk: top k predicted EC numbers
#-mode p: prediction mode, predict EC numbers only
#-mode r: recommendation mode, recommend EC numbers with predicted probabilities, the higher the better
#-mode h: hybird mode, use prediction, recommendation and sequence alignment methods

To re-implement our experiments or offline use, pls use read the details below:

Prerequisites

• Python >= 3.6
• Sklearn
• Xgboost
• conda
• jupyter lab
• ...

Create conda env use env.yaml

git clone git@github.com:kingstdio/ECRECer.git
conda env create -f env.yaml

Preprocessing

Download and prepare the data set use the.

prepare_task_dataset.ipynb

Or directly download the preprocessed data from aws public dataset and put it in the rootfolder/data/datasets/

High throughput benchmarking

Train

python benchmark_train.py

Test

python benchmark_test.py

Evaluation

python benchmark_evaluation.py

Production

python production.py -i input_fasta_file -o output_tsv_file -mode [p|r] -topk 5

Citations

If you find these methods valuable for your research, we kindly request that you reference the pertinent paper:

@article{shi2023enzyme,
  title={Enzyme Commission Number Prediction and Benchmarking with Hierarchical Dual-core Multitask Learning Framework},
  author={Shi, Zhenkun and Deng, Rui and Yuan, Qianqian and Mao, Zhitao and Wang, Ruoyu and Li, Haoran and Liao, Xiaoping and Ma, Hongwu},
  journal={Research},
  year={2023},
  publisher={AAAS}
}

Stargazers over time