Depends on the use case. If one is trying to broadly sequence a genome or exome, all of that is correct. For something like a liquid biopsy test that's looking

Depends on the use case. If one is trying to broadly sequence a genome or exome, all of that is correct. For something like a liquid biopsy test that's looking for circulating tumor DNA, a given mutation may only exist in 0.5% of the reads that cover a particular locus. Then you have to deduplicate to tell which are PCR duplicates of the same original strand, and which have different progenitor fragments. Low-coverage long reads are bad for this.