Where the data comes from
Nothing is vendored in the repository — every run fetches the current data. The three ONT BAMs and the PacBio Iso-Seq BAM total about 159 GB and are never downloaded whole; htslib range-reads only the blocks covering the vaccine-gene windows.
What the portal actually holds
Derived from osteosarc.com's own file manifest, not asserted — so "isn't there PacBio for T2?" has an answer anyone can re-derive. Counts are files whose path names a platform, so they include intermediates and indexes, not just primary data.
Reproducing a run
Every command, in order, from an empty directory. Each block is copyable.
How it was run
Every option, measured against the canonical pipeline. Anything not marked stock was changed deliberately, and says why.
Where it ran, and with what
Compute
The pipeline runs on GitHub-hosted ubuntu-latest runners, with
the sequencing samples as a parallel matrix — one job each, because a single
job doing them all would consume most of GitHub's six-hour ceiling. The leg
is the sample, not the biopsy: T1 was sequenced on ONT and again on PacBio,
and a matrix keyed on the timepoint would have to drop one of them. It fires
on a schedule every Monday at 04:00 UTC, on a manual dispatch (which
takes an Exacto version as input), and on any push that touches the pipeline.