Oooh quantum century!

Yeah we definitely have some cases of these foldback reads, this is useful because these reads need some QC'ing but technically ~100% of the sequence aligns so they look as good as any other SV-containing read with sup alignments.

Highly repetitive is easier to eyeball. We have some 700 Kb reads where 400 bp (yes bp) align to the reference the rest is soft-clipped "TGTGTG...". Some unaligned reads are legitimately just telomeres (the cattle reference is incomplete), so hard to find a single metric to separate real from fake

Sadly have to decide between maximising the throughput or hitting the Euler hard quota with uncompressed genomes. Painful to see how much time it adds though

Compared to `mash triangle` for 208 bgzipp'd cattle genomes on 4 threads, it only took 65 min vs 230 min. Used slightly more RAM (8.2 GB vs 3.3 GB), but easily worth the speedup! Mash defaults are also k=21 and 1k sketches, so not a fair comparison to k=31 and 10k (not sure about sketches v buckets)

No longer need to guess what the dorado bam tags mean!

Any now his watch is ended.

Hopefully there will be a point release or something to get IGV loading cram3.1

Hopefully htsjdk will have merged support for CRAM 3.1 around the same time. Recompressing 3.1 to 3.0/BAM for collaborators so they can view the alignments in IGV is a bit of a pain, and 3.1 being the default will probably lead to that issue more often.