/DeePathology

Tissue and Cancer Type Identification using Deep Neural Networks

Primary LanguagePython

DeePathology

Deep Multi-Task Learning for Inferring Molecular Pathology from Cancer Transcriptome

Authors: Behrooz Azarkhalili, Ali Saberi, Hamidreza Chitsaz, Ali Sharifi-Zarchi

All codes for obtaining and pre-processing data are provided in PreProcess folder. These are custom R scripts for downloading mRNA and miRNA expression profiles from The Cancer Genome Atlas (TCGA) from Genomic Data Commons (http://gdc.cancer.gov).

All codes for hyper-parameter optimization and training different deep neural network architectures (Variational AutoEncoder, Contractive AutoEncoder, Dropout Variational AutoEncoder and Variational Contractive AutoEncoder) are provided in DeepLearning folder.

Deep neural networks are developed using Keras python package and TensorFlow backend.