Fig. 1From: Cancer type prediction based on copy number aberration and chromatin 3D structure with convolutional neural networksPerformances of our proposed method with different design options. a With different HiC data configurations. From left to right: baseline model (2D CNN); baseline with hESC only; baseline with IMR90 only; baseline with both types of HiC data. The last configuration leads to the optimal performance. b With different network and HiC combinations. From left to right: 1D CNN without HiC data; 1D CNN with HiC data; 2D CNN without HiC data; 2D CNN with HiC data. The last configuration leads to the optimal performanceBack to article page