Fig. 6From: An efficient scRNA-seq dropout imputation method using graph attention networka–c show the visualizations of the raw matrix, noised matrix and denoised matrix after GNNImpute. d, e show the ARI and NMI of different methods. f, g show the heat maps of the raw expression matrix and noised matrix. h– m show the heat maps of the expression matrix imputed by GNNImpute and the other five methodsBack to article page