Fig. 2

Comparison of the runtimes and memory usage of common CRISPR target discovery tools over an increasing number of targets and permitted mismatches. Five random CRISPR guide sets were run for each target-count (x-axis) and permitted mismatch level (y-axis). Plotted are the mean runtime with standard deviation bars for each set of replicates. a Running time per sequence for increasing numbers of target sites and b their corresponding memory usage. FlashFry benefits from aggregating all guide-to-genome comparisons in one pass of the database, matching BWA’s performance at hundreds of targets for five mismatches, and thousands of targets at four mismatches. Only BWA and FlashFry were run for the 10,000 and 100,000 target searches