APPM:Neoantigen Presentation Prediction Model
Neoantigen Presentation Prediction Model, named APPM, is a neoantigen predictor build by HLA-peptides mass spectrometry data and convolutional neural network (CNN). Compared to the netMHCpan4.0, our framework demonstrates higher values of area under the ROC curve (AUC) in some HLA alleles.
Dependencies
tensorflow-gpu 1.14
conda install --channel https://conda.anaconda.org/fwaters tensorflow-gpu==1.14
pandas
sklearn
Usage
Our models (20 availbale alleles) have been trained. If just for prediction neoantigens, you can run the prediction.py directly.
python script/prediction.py [HLA allele] [intput_file] #DATA/predict_data/
python script/prediction.py A0101 pep9
If you want to retrain, you can run APPM.py.
python script/APPM.py 0 [HLA allele]
python script/APPM.py 0 A0101
If you want to continue training, you can run APPM.py.
python script/APPM.py 1 [HLA allele]
python script/APPM.py 1 A0101