Figure 2. Relative Root Mean Square Error (RRMSE) for every dataset of read length 150 bp grouped by software pipeline.

For each pipeline, datasets tend to group around an RRMSE of 0.5 or less, with notable exceptions: Taxator-tk, MetaPhlAn and Constrains, with the two latter cases having a number of datasets that could not be represented here. Sigma also has a few outliers.