adler-sudo / dnnPreMiR

identify pre-miRNAs using CNN models trained on huamn pre-miRNAs dataset

###################################################################################### Using Trained CNN model to predict pre-miRNAs. Data are from the human pre-miRNAs and hairpin structure sequences from human coding regions.

######################################################################################

USAGE: python isPreMiR.py -s RNAsequence for example: python isPreMiR.py -s CUCCGGUGCCUACUGAGCUGAUAUCAGUUCUCAUUUUACACACUGGCUCAGUUCAGCAGGAACAGGA

      python isPreMiR.py -i inputFilePath [-o outputFilePath]

(note:The length of RNA should be not more than 180, all the input RNA sequence will be truncated or padded into 180 length)

Requirements

• Python 3
• Tensorflow > 0.12
• Numpy
• RNAfold (/bin/RNAfold)

Training

We trained the CNN model,RNN model and the concat model to train the dataset.

Evaluating

Evaluate the trained model on test dataset.