Fig. 2

Absolute difference between sparsity in the ground truth and in the different pipelines for the three simulated datasets. (The lower the better). Results are aggregated according to the imputation method used in the pipelines and are reported as mean values calculated across the six pipelines using the given zero-imputation tool; error bars represent the corresponding standard deviations. “None” identifies pipelines where no zero-imputation step was performed, i.e. normalization-only pipelines