Skip to main content
Fig. 4 | BMC Medical Genomics

Fig. 4

From: Convolutional neural network models for cancer type prediction based on gene expression

Fig. 4

Interpretation of the 1D-CNN model. a Distributions of gene-effect scores for individual cancer and normal classes. Colors correspond to cancer types denoted in Fig. 4b. b t-SNE plots of pan-cancer and normal samples by expression of marker genes identified using different thresholds. c Marker genes identified in each class with a criterion of gene-effect score > 0.5. The dashed line denotes the average number of marker genes identified across 34 classes. d-e Differential expression of marker genes and other genes between sample classes. Here differential expression is presented by an absolute difference between a class (normal or BRCA) and all other samples in log2(FPKM+ 1). f Pan-classes gene-effect scores of three marker genes of BRCA. g Functions associated with marker genes identified in each class

Back to article page