Read cell labels produced by another tool

Accepts a CSV carrying one row per cell with a sample identifier, the event's index within that sample's file, and a label. Column names are matched loosely, so a table exported from a clustering package usually needs no editing.

Usage

read_external_labels(
  path,
  sample_col = NULL,
  event_col = NULL,
  label_col = NULL
)

Arguments

path

CSV path

sample_col, event_col, label_col

column names; NULL matches by convention

Value

data.frame(sample_id, event_index, label)

Details

WHY THE KEY MUST BE EXPLICIT AND NOT POSITIONAL. Tools subsample. cyCONDOR takes max_cell events per file and cyRAVEN takes its own cap for the embedding, so row i of one table is not row i of the other and a positional join silently mislabels every cell. Requiring the event index makes the mismatch an error instead of a result.