The Keras implement of MAM.
The Keras implement of In silico design of MHC class I high binding affinity peptides through motifs activation map.It is a framewrok of both the unknown MHC-I peptide prediction and novel peptide generation. Our paper is still under reviewed.
Prerequisites: Python 3.5.3 Tensorflow 1.4.0 Scipy 0.19.1 Numpy 1.13.3 Pandas 0.20.3 Sklearn 0.18.2 Keras 2.1.3 Gensim 3.2.0
Getting Started: The pipeline is as follows:
- Amino Acid to Vector: Convert the Amino Acid into n-dimemsion vector.
I. Find config.ini and Set Vec == True
II. Run python main.py config.ini
- Train model.
I. Find config.ini and Set train == True
II. Run python main.py config.ini
- Evaluate model. Do the prediction on the test dataset.
I. Find config.ini and Set evaluate == True
II. Run python main.py config.ini
- Inference model. Do the prediction on the test dataset as well as generation.
I. Find config.ini and Set Inference == True
II. Run python main.py config.ini
- Fine-tuning. The setting of train, evaluate and inference model are the same operation as we mentioned above.
Dataset The training dataset is available in train_data file while the test dataset is in test_data file. Moreover, the label name in drawing function (e.g., heatmap, tsne) need to be rewritten by the user.
Acknowledge: Some of the code function refer from HLA-CNN https://github.com/uci-cbcl/HLA-bind. We appreciate so much for their excellent code.
If you have any question you can email runshengyu@gmail.com for help.