These functions modify the metadata of either an ir_aggregate_isofiles()
result (ir_aggregated_data) or a collection of isofiles read with
ir_read_isofiles() (ir_isofiles).
Usage
ir_filter_metadata(isofiles, ...)
ir_mutate_metadata(isofiles, ...)
ir_join_metadata(isofiles, y, by)Arguments
- isofiles
datasets aggregated from
ir_aggregate_isofiles()(ir_aggregated_data) or a collection of isofiles fromir_read_isofiles()(ir_isofiles)- ...
passed to
dplyr::filter(),dplyr::mutate(), ordplyr::left_join()respectively- y
data frame to join to the metadata
- by
character vector of columns to join by (passed to
dplyr::left_join())
Details
For ir_aggregated_data, the operation is applied once to the combined
$metadata data frame. For ir_filter_metadata(), the filter then cascades
to all other datasets: traces, cycles, and scans are filtered by the
remaining uidx + analysis combinations; resistors and problems are
filtered by the remaining uidx values.
For ir_isofiles, the same operation is instead applied individually to
each row (i.e. to each file's own nested datasets), since an ir_isofiles
object has no combined metadata to operate on. Within each row, the filter
cascade uses whichever linking columns are present (typically analysis).
For ir_filter_metadata(), any file whose metadata ends up with 0 rows after
the filter is removed from the ir_isofiles collection entirely.
Operating on an unaggregated ir_isofiles object is supported for convenience,
but is significantly slower than operating on an ir_aggregated_data
result, because the operation has to be carried out separately on every file
rather than once on the combined metadata. For anything beyond small
collections, prefer aggregating first with ir_aggregate_isofiles() and then
applying these functions to the result.
After filtering, columns that are entirely NA across all remaining rows are
dropped from every (non-empty) dataset.
All three functions also clear the not-aggregated column information (columns
present in the source files but not included in the aggregator) from every
dataset, since that information is no longer meaningful after the metadata has
been modified.