List the measurement timepoints entering each growth-rate window

Reports which durations actually go into the fit for every period, so that a report or a log can state the measurements it used rather than only the window bounds. The bounds alone are ambiguous: whether a timepoint sitting exactly on a boundary is included depends on the window convention, and a window may silently contain far fewer points than intended.

Usage

get_period_timepoints(se, periods, lfc_assay = NULL)

Arguments

se

SummarizedExperiment with lfc_assay.

periods

named list of two-element numeric windows, as passed to compute_growth_rates.

lfc_assay

string; assay to read durations from. NULL uses the default input assay of the "time-course" fit profile.

Value

data.table with one row per period: period, window (the bounds as "start - end"), n (number of distinct timepoints) and timepoints (those timepoints, sorted, comma-separated). Periods matching no measurement yield n = 0.

timepoints is the union of the durations present anywhere in se, not a per-plate or per-well listing. On a uniform sampling grid the two coincide; if plates were read on different schedules, a duration contributed by only one barcode still appears here, and n is therefore an upper bound on what any single well contributed to the fit.

Details

Uses the same half-open [start, end) rule as compute_growth_rates, so the answer always matches what was fitted.

See also

compute_growth_rates, fit_SE.timecourse

Examples

if (FALSE) { # \dontrun{
get_period_timepoints(se_tc, list(early = c(44, 92), late = c(92, 140)))
} # }