Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Deep learning and multi-omics approach to predict drug responses in cancer

Fig. 2

Deep neural network architecture combining different omics data. Gene expressions, CNV, and Gene mutations data are fed into a graph embedding layer, respectively, whereas RPPA and metabolomics data are fed into a dense embedding layer, respectively. In a graph embedding layer, information about interactions among genes obtained from HINT database is incorporated, and only genes with mutual interactions are distilled. Then a dense layer is applied for further learning the latent features of each omics dataset. Eventually, an attention layer is adapted to predict the final drug responses with distinct attention to different omics features

Back to article page