Deep6mA is a deep-learning-based framework to predict 6mA-containing sequences.

• Python3

• numpy==1.14.2

• torch==0.4.0a0+0e24630 ( pytorch )

• scikit_learn==0.19.1

• data: data of rice for model training ( data of Arabidopsis thaliana, Fragaria vesca and Rosa chinensis for validation on other three species)

• demo_test:

  • test.fa: data for testing scripts
  • expected results
    • predout.tsv: prediciton results

• result:

  • trained_models: trained models for rice(5-fold)

The script main_train.py is used to train model. The required arguments

• model_type: cnn-rnn
• pos_fa/neg_fa: the fasta file for positive samples/negative samples ( the length of sequences should be no more than 41bp)
• out_dir: the path of output directory

This script ouput the trained model and prediction result in the out_dir.

python main_train.py -m model_type -pos_fa pos_fa -neg_fa neg_fa -od out_dir

The script main_test.py is used to predict if a given sequence contain 6mA sites. The required arguments

• model_type: cnn-rnn
• input_fa: a fasta file for test samples ( the length of sequences should be no more than 41bp)
• model_dir: the path of model directory
• out_fn: output file

This script ouput the prediction scores for given sequences.

python main_test.py -m model_type -in_fa input_fa -md model_dir -outfn out_fn