Fig. 6

Additional experiments with varying k and (w, k) for the HIBF. The maximum false-positive rate for the (H)IBF was fixed to \(1.5\%\), and we used \(t_{max}=192\) for the HIBF. a For an HIBF using canonical k-mers, the choice of k had little impact on index size as well as accuracy and achieved the best overall accuracy. Compressing the index with minimizers decreased the accuracy. Although generally all tools achieve excellent accuracy, with a minimum of \(99.5\%\) (Metagraph), it has to be noted that the accuracy is biased by a high number of true negatives. Thus, small differences in accuracy have a high impact on the actual numbers of false positives and negatives. b The index size in GiB. c The query time in seconds of the experiments on the RefSeq data set