Current position in this corpus

Long-read WGS is not uniformly better for every task, but it is consistently better where read length, phasing, or native-molecule information matters.

Where long reads clearly help

  • insertions larger than roughly 10 bp
  • structural variants in repetitive regions
  • hard medically relevant genes and sequencing dead zones
  • breakpoint-level interpretation of complex rearrangements
  • phasing of compound events
  • native methylation-aware diagnostics

Where short reads remain competitive

  • many SNVs outside repetitive regions
  • many deletions in nonrepetitive regions
  • cost-efficient high-throughput cohort sequencing
  • mature tooling and established clinical workflows

Supporting sources

Main tension

  • One view in the corpus favors targeted deployment where long reads have the highest incremental value.
  • Another view favors a unified long-read clinical platform that replaces multiple follow-up assays.

Open questions

  • At what cost and coverage point does long-read WGS become the default rather than the escalation path?
  • How much of the remaining short-read advantage is technical, and how much is simply ecosystem maturity?