Figure 2

Denoising “misses” with FlowClus. As FlowClus denoises flowgrams by comparing pairs of flow values, it records the flow values of the cluster and the read each time they are judged as being distinct. The set of these “misses” for a denoising run can be visualized as a levelplot, such as those shown here. The red-orange colors represent a large number of misses at those particular pairs of cluster and read flow values. A: Denoising using a constant distance value of 0.50. B: Denoising using a multiple of five distances based on the standard deviations of Balzer et al. [22].