important! Very sorry, some data in the additional file is wrong due to table format problem. We update the new version in the ./additional_file.pdf
This repository contains code for "iDNA-ABF: a deep learning sequence modeling framework for DNA methylation prediction".
Here is the article link and webserver link
We will provide a google drive for download dataset and model parameters in the future.
We have provided a base parameters by One drive, you can download by this One drive share
Any question is welcomed to be asked by issue and I will try my best to solve your problems.
Thanks to Yingying Yu (She used to be a member of Weilab and now continues her phd life in CityU). She offers a nni version based on pytorchlighting and you can reproduce relevant results by her repository.
You can change parameters in configuration/config.py to train models.
You can change model structure in model/ClassificationDNAbert.py to train models.
You can change training process and dataset process in frame/ModelManager.py and frame/DataManager.py .
Besides, dataset in paper "iDNA_ABF" is also included in data/DNA_MS.
You should download pretrain model from relevant github repository.
For example, if you want to use DNAbert, you need to put them into the pretrain folder and rename the relevant choice in the model.
python main/train.py