Fig. 1

Illustration of sequencing settings that pose challenges for count-based log-enrichment approaches. “Ideal”: in an ideal world, we would always have high read coverage, with reads long enough to span the region of interest (length L ). “Short reads”: in the short read setting, short, possibly overlapping reads each only cover a portion of the region of interest. For illustrative purposes, we have displayed short reads in different positions and color-coded by sequence identity, but in general, positional information would not be known and would have to be deduced through read mapping or similar. “Sparse reads”: in the sparse read setting, even if the reads are sufficiently long to cover the region of interest, few are available, making cLE estimates high variance. “Hybrid reads”: in the hybrid read setting, long- and short-read datasets are combined into a single sequencing dataset. Consequently, this setting is plagued with problems that arise in both short and sparse settings, although the combined types of reads should in principle be useful