Fig. 1From: DEEPSEN: a convolutional neural network based method for super-enhancer predictionThe pipeline of DEEPSEN. The data we used were from GEO. Firstly, we do data preprocessing and feature calculation. Secondly, we construct three models with different numbers of convolutional layers and train them. Thirdly, we evaluate Pearson correlation coefficient to rank the features for predicting super-enhancers. Finally, we do performance evaluation and analysisBack to article page