Comments on revised manuscript

In this revision, the authors answered some of my comments. My primary concern from the first version of the manuscript is whether the authors use the best possible de novo assemblers. They compare different technologies, and their conclusion relies on achieved results. In their previous paper, they showed that Canu consistently produces a much longer final sequence than other solutions. From the manuscript, it is not clear if they use methods for the removal of haplotypic duplications (ie. purge duplication). I deem that they should test at least one another assembler (ie. Flye) for error-prone reads. Flye is more resilient to haplotypic duplications. Similarly, IPA is rarely used for hifi reads, and most of the authors use hifiasm. Even PacBio uses hifiasm in their analysis. The newest version of Flye and hifiasm are fast assemblers, so I deem their usage will not require significant computational resources. From above, I argue that the authors need to provide more results to support their claims.