The problem. Even with long reads, some regions resist assembly: the human X chromosome’s centromere is a vast array of near-identical repeats where reads can’t be uniquely placed and assemblies fragment. Spanning it — walking from one side to the other without breaks — was a standing challenge that defined “unfinished” in the reference.

The idea. Jain and colleagues applied the very longest nanopore reads to traverse the centromeric satellite array of the X chromosome, producing a contiguous, linear assembly through a region short reads and even ordinary long reads couldn’t resolve. The key ingredient is read length pushed to its extreme — reads long enough to anchor across the repeat.

Why it matters. This is the concrete demonstration of what the previous paper promised: not just more sequence, but finishing the parts that were structurally impossible before. It’s the direct lineage to the complete T2T genome later on this list. For me, it draws the sharpest possible line around short-read variant calling — there are entire chromosomal features my variant_calling_nf reference can’t even represent, which only assemblies like this fill in.

Verdict. Foundational as a milestone toward the complete human genome. Read it paired with the NA12878 nanopore paper — sequencing the genome, then conquering its single hardest region — as the two-step case for why long reads changed what “reference” means.