Spillover spreading matrix from single-stain controls — spillover_spreading

The published measure (Nguyen et al. 2013): for a single-stain control of fluorochrome i, split into its positive population P and negative population N on its primary detector, the spreading coefficient into detector j is

Usage

spillover_spreading_matrix(stains, min_events = 200L)

Arguments

stains: named list of reads from read_fcs_resolved(), compensated.
min_events: fewest events each of P and N must hold. Default 200.

Value

data.frame, one row per control and receiving detector: control, source_detector, receiver_detector, n_positive, n_negative, delta_median, spread_positive, spread_negative, ss. NULL when no control can be used.

Details

SS_ij = sqrt(max(0, sigma(P, j)^2 - sigma(N, j)^2)) / sqrt(F(P, i) - F(N, i))

where sigma is the robust standard deviation (84.13th minus 50th percentile) of the compensated values and F the median. The square root of the intensity difference normalises out how bright the control was, so the coefficient is a property of the fluorochrome-detector pair on this instrument, comparable with published values and between panels.

The primary detector of each control is the one where it is brightest (the highest 99th percentile); the positive and negative populations are split at the density minimum on that detector, on an arcsinh scale with cofactor 150, or at the median where none is found.