LSTM for binary sequence classification
- kmer based SVM/RF approach: https://github.com/philippmuench/plasmidminer
structure of training data (see cami_10k.csv)
sequence,target
ACGTAGCT,1
ACCCTAAG,1
TCGTAACG,0
ACTGACCG,0
...
sudo apt-get install graphviz
virtualenv env
source env/bin/activate
pip install matplotlib keras h5py tensorflow numpy pandas pydot-ng pydot graphviz
python train.py
final model will saved to model.json and model.h5 plus plots accuracy.png and loss.png
this script will use the model generated in the training process to predict list of sequences found in test.csv
python predict.py