D3Impute: Dropout-aware discrimination, distribution-aware modeling, and density-guide imputation for scRNA-seq data
Fig 3
Benchmarking results under a 60% dropout rate.
(A) 2D visualization of the masked dataset (seed = 55) before and after imputation, illustrating the restoration of data structure. (B) Circular quadrant plot summarizing the performance of multiple imputation algorithms across five random masking trials (seeds = 44, 55, 66, 77, 88). The circle is divided into four quadrants, each representing one evaluation metric: Pearson correlation coefficient (PCC), root mean square error (RMSE), proportion of correctly recovered non-biological zeros, and proportion of falsely imputed biological zeros. Within each quadrant, the bars show the mean performance of each algorithm over five random seeds, and the error bars indicate the corresponding standard deviation (mean SD).