Unsupervised discovery over every eligible channel, writing into
<outdir>/explore/. Every argument beyond reads, fpr, opt and outdir
is optional: supplied, it makes the output better; absent, the run degrades to
what a standalone clusterer would produce and records that it did so in
explore_provenance.csv.
Usage
run_explore(
reads,
fpr,
opt,
outdir,
transforms = NULL,
gates = NULL,
verdicts = NULL,
pops = NULL,
group_of = NULL,
confounding = NULL,
file_paths = character(0),
total_counts = NULL
)Arguments
- reads
Named list of reads.
- fpr
Panel resolution.
- opt
Option list.
- outdir
Run output directory.
- transforms
Per-panel transforms. Derived here if absent.
- gates
Per-sample gate objects, for thresholds and the viability marker.
- verdicts
Per-sample staining QC verdicts. Recorded, NOT used to exclude: explore does not depend on the CD45 gate, so a sample the declared path had to drop still contributes here. That is a capability rather than an oversight, which is why it becomes a column instead of a filter.
- pops
Per-sample scored populations, for the bridge tables.
- group_of
Named character vector, sample id -> group.
- confounding
One-row data frame from the batch/group confounding check.
- file_paths
Full paths of the FCS files the run was given. Needed only when the declared pipeline has already freed the event matrices, because
reads[[s]]records a basename and cannot be re-opened on its own.