This repository includes the implementation of 'MR-DNA: Flexible DNA 5mC-Methylation-Site Recognization in DNA sequences using Token Classification'.
Download MR-DNA from the github repository.
git clone https://github.com/husonlab/MR-DNA.git
cd MR-DNA
We recommend you run MR-DNA in a Python virtual environment built by Anaconda, creating a conda environment equipped with the required packages from the MR-DNA yml file.
conda env create -n MR-DNA --file MR-DNA.yml
# If failed, please try to update your conda to the latest version using
conda update -n base -c conda-forge --all
# Check the installation of the required environment
conda info --env
conda activate MR-DNA
# Install the required package additionally
pip install pytorch-crf
Training DNA-MR on DNA-MR-50 dataset
cd MR-DNA
python ./scripts/main.py --dataset MR-DNA-50 --status train --model DistilBertCRF_MethyLoss --savePath result/model/
Evaluating DNA-MR performance on the test dataset. Training steps saved models whose performance improved. The user needs to give the specific model name when runing the following command.
cd MR-DNA
python ./scripts/main.py --dataset MR-DNA-50 --status test --model DistilBertCRF_MethyLoss --savePath result/model/ --trainedModel <model>
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent