Skip to contents

Density gating of intensity values in marker for each unique subset of subset_col

Source: R/getDensityGates.R
getDensityGates.Rd

For each unique value in subset_col, gate using density and estimated derivatives to identify cutoff at shoulder (i.e., point of tapering off) relative to the peak for marker (intensity values). The strategy of cutting at the shoulder mimics the strategy to gate relative to a unimodal background negative subpopulation, which is capable of capturing dim subpopulations.

Usage

getDensityGates(
  intens_dat,
  marker,
  subset_col,
  bin_n = 512,
  peak_detect_ratio = 10,
  pos_peak_threshold = 1800,
  neg_intensity_threshold = -1000
)

Arguments

intens_dat

dataframe of pre-gated (compensated, biexp. transf, openCyto steps) intensity values where cols=intensity value per marker, rows=each sample

marker

string for the marker(s) to gate on the names need to match exactly the column name in intens_dat

subset_col

string for the column name to indicate the subsets to apply density gating on will perform operation on subsets corresponding to each unique value in column

bin_n

numeric to be passed to n parameter of density(n=bin_n) for number of equally spaced points at which the density is to be estimated
Default is 512, which is the default of density(n=512)

peak_detect_ratio

numeric threshold for eliminating small peaks where a peak that is < than the highest peak by peak_detect_ratio times will be ignored
Default=10

pos_peak_threshold

either:

  • numeric for threshold to identify a positive peak for all or

  • a dataframe if supplying multiple marker to gate. The dataframe needs to be supplied with 2 columns named marker and pos_peak_threshold and rows for the marker to gate

Default is 1800 (note this is on the biexponential scale) for all marker

neg_intensity_threshold

numeric for threshold to filter out any "very negatively" expressed cells in the density estimation to avoid over-compression and difficulty in distinguishing peaks and the gates
This is only applied as a filter for the density estimation, the cells < neg_intensity_threshold are retained in the intensity matrix for other steps
Expects the neg_intensity_threshold is on the same scale as the transformed data in intens_dat
Default is NULL: no filters applied and density estimation based on all cells in corresponding subsets.
Suggested for biexp. transformed data is -1000 which corresponds to ~-3300 on the original intensity scale)

Value

tibble of gates/cutoffs for marker for each unique subset found in subset_col where

  • rows correspond to unique values in subset_col

  • , columns correspond tomarker

Examples

# Create a fake dataset
set.seed(100)
intens_dat<-tibble::tibble(
               CD3_pos=rep(c(0, 1), each=50),
               CD4=rnorm(100, 100, 10),
               CD8=rnorm(100, 100, 10)
)

# Run density gating, leaving other params at suggested defaults
# number of bins suggested is 40 but default is at `bin_n=512`,
# which is the default for the R base density() function
getDensityGates(intens_dat, marker="CD4", subset_col="CD3_pos", bin_n=40)
#> # A tibble: 2 × 2
#>   CD3_pos   CD4
#>     <dbl> <dbl>
#> 1       0  88.9
#> 2       1 113.