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