Fig. 2
From: ENGEP: advancing spatial transcriptomics with accurate unmeasured gene expression prediction
Prediction comparison for MERFISH. a Performance comparison between ENGEP and five benchmarked methods via fivefold cross-validation. The scatter plots show the PCC values of every gene across ENGEP (y-axis) and the benchmarked methods (x-axis). The red line is the \(y=x\) line. The AR value greater than 1 indicates that ENGEP predicts better than others on more genes, while the P value (Wilcoxon rank-sum test) shows the significant difference between correlations across two methods. b Measured (top) and predicted (bottom) expression patterns of test genes selected with different PCC values (bottom arrow). c Negative correlation between the PCC values and sparsity values of each gene. d The ability of ENGEP to correct for low-quality genes. The predicted patterns (middle) differ from the measured patterns (top) but are in agreement with the ISH images (bottom)