Fig. 1
From: A pan-cancer somatic mutation embedding using autoencoders
Model architecture. Scheme of the multi-modal autoencoder architecture for both deleterious and non-deleterious mutational profiles. Input and output dimension have 12424 genes. The encoder and decoder functions contain one hidden layer each of 400 activation functions (neurons). The latent layer of each autoencoder has 50 activation functions. Highlighted in red is the latent space Ltot which contains signal from both types of mutational profiles