Deep Learning model to identify novel anticancer molecules in food sources

We plan to use a deep neural network model we already created and trained to identify novel anti cancer peptides. We trained the model by encoding physicochemical properties of various peptides, or short amino acid sequences, to identify peptides that are Anti-cancer vs. non-anti cancer. We trained the model on a custom dataset of ~1600 peptides, and the model achieved a near perfect accuracy of 98% in identifying certain peptides. We plan to use this model and to potentially retrain it with different neural network architectures to identify novel anticancer peptides in various food sources. This is important to potentially identify novel superfoods with anticancer properties. We plan to work on this by implementing novel architectures such as Convolutional neural network or recurrent neural network. We also might try to use generative adversial networks to create new anticancer molecules based on sequential information. To identify new potential superfoods, we will have our model scan through a database of peptides in various food sources, such as FermfoodDB.