Epitope Generation for Peptide-based Cancer Vaccine using Goal-directed Wasserstein Generative Adversarial Network with Gradient Penalty
Keywords: Convolutional neural network, deep learning, generative adversarial network, immunogenicity, peptide-based cancer vaccine
Abstract: We introduce a novel immunogenicity goal-directed peptide sequence generator in a Wasserstein generative adversarial network (GAN) with gradient penalty. The GAN is trained using the bladder cancer epitope sequences that are predicted to bind with the human leukocyte antigen, HLA-A*0201, to trigger cytotoxic T-cell immune responses. The convolutional neural network-based generator is guided by an immunogenicity predictor from DeepImmuno-CNN and a critic to generate immunogenic epitopes for bladder cancer vaccines. The convolutional neural network-based immunogenicity predictor is trained with class I peptide human leukocyte antigen sequences from the immune epitope database to produce a continuous immunogenic score. We incorporated the trained immunogenicity predictor by training the generator with the predicted immunogenicity score of the generated peptide sequences. We showed our generator can produce more immunogenic peptides after adding the predictor and can produce peptides that are similar to the epitopes shown in bladder cancerous cells.
Submission Number: 7
Loading