Fig. 3

Error counts for the different algorithms after comparing with simulated data on different scenarios. Accessory genome inflation refers to the number of erroneous clusters that do not correspond to any simulated gene cluster. Missing genes refer to false-negative gene calls where the annotation is not present in the final pangenome. Even in simulations of pangenome variation from a single E. coli reference with only relatively simple sources of error, a Panaroo outperforms other methods across a variety of gene gain/loss rates and mutation rates. In more realistic simulations of sequencing data, b the only method with reasonable control of the error rate is Panaroo