Skip to contents

fit_drug_response_metrics

Source: R/fit_utils.R
fit_drug_response_metrics.Rd

Reference fit function replicating standard gDR Hill curve fitting (3-parameter model, x_0 fixed at 1). Produces output column-compatible with fit_SE and logisticFit, including p_value, rss, x_AOC_range, x_max, and x_sd_avg. For use with apply_fit_to_se or apply_fit on single-agent or time-course data.

Usage

fit_drug_response_metrics(
  avg_dt,
  x_col = "x",
  capping_fold = 5,
  range_conc = c(0.005, 5),
  pcutoff = 0.05,
  n_point_cutoff = 4L,
  force_fit = FALSE,
  cap = 0.1
)

Arguments

avg_dt

data.table of averaged data for one (drug x cell line x normalization_type) triplet. Should contain columns Concentration, the response column named by x_col, normalization_type, and optionally x_std (used for x_sd_avg).

x_col

string; name of the response column in avg_dt. Defaults to "x" (the standard Averaged assay column name). Set to e.g. "NormalizedGrowthRate" when fitting time-course growth-rate data.

capping_fold

numeric capping fold passed to cap_xc50

range_conc

numeric vector of length 2 specifying the concentration range used to compute x_AOC_range. Default c(5e-3, 5), matching the fit_SE() default.

pcutoff

numeric p-value cutoff for the fit. Default 0.05.

n_point_cutoff

integer minimum number of concentration points required to attempt fitting. Default 4L.

force_fit

logical; if TRUE, force fitting even when the number of points is below n_point_cutoff. Default FALSE.

cap

numeric; upper cap on xc50 relative to the concentration range. Default 0.1.

Value

Named list of fit metrics fully compatible with the standard gDR Metrics assay schema: ec50, xc50, h, x_inf, x_0, r2, rss, p_value, x_mean, x_AOC, x_AOC_range, x_max, x_sd_avg, N_conc, maxlog10Concentration, fit_type, normalization_type.

See also

fit_drug_response_metrics_4p for a 4-parameter variant that fits x_0 freely.

Examples

dt <- data.table::data.table(
  Concentration = c(0.001, 0.01, 0.1, 1, 10),
  x = c(0.95, 0.8, 0.5, 0.2, 0.1),
  normalization_type = "RV"
)
fit_drug_response_metrics(dt)
#> $normalization_type
#> [1] "RV"
#> 
#> $x_mean
#> [1] 0.5070152
#> 
#> $x_AOC
#> [1] 0.4929848
#> 
#> $x_AOC_range
#> [1] 0.5471268
#> 
#> $x_max
#> [1] 0.1
#> 
#> $x_sd_avg
#> [1] NaN
#> 
#> $N_conc
#> [1] 5
#> 
#> $maxlog10Concentration
#> [1] 1
#> 
#> $ec50
#> [1] 0.07913255
#> 
#> $xc50
#> [1] 0.09530136
#> 
#> $h
#> h:(Intercept) 
#>     0.6523032 
#> 
#> $r2
#> [1] 0.9996676
#> 
#> $rss
#> [1] 0.00018015
#> 
#> $p_value
#> [1] 0.000006059718
#> 
#> $x_0
#> [1] 1
#> 
#> $x_inf
#> [1] 0.05710528
#> 
#> $fit_type
#> [1] "DRC3pHillFitModelFixS0"
#> 

# Custom response column (e.g. time-course growth rates):
dt2 <- data.table::data.table(
  Concentration = c(0.001, 0.01, 0.1, 1, 10),
  NormalizedGrowthRate = c(0.98, 0.82, 0.55, 0.21, 0.09),
  normalization_type = "GR"
)
fit_drug_response_metrics(dt2, x_col = "NormalizedGrowthRate")
#> $normalization_type
#> [1] "GR"
#> 
#> $x_mean
#> [1] 0.5299304
#> 
#> $x_AOC
#> [1] 0.4700696
#> 
#> $x_AOC_range
#> [1] 0.5198366
#> 
#> $x_max
#> [1] 0.09
#> 
#> $x_sd_avg
#> [1] NaN
#> 
#> $N_conc
#> [1] 5
#> 
#> $maxlog10Concentration
#> [1] 1
#> 
#> $ec50
#> [1] 0.111215
#> 
#> $xc50
#> [1] 0.1259556
#> 
#> $h
#> h:(Intercept) 
#>     0.6665989 
#> 
#> $r2
#> [1] 0.9981679
#> 
#> $rss
#> [1] 0.001068109
#> 
#> $p_value
#> [1] 0.00007841885
#> 
#> $x_0
#> [1] 1
#> 
#> $x_inf
#> [1] 0.03980943
#> 
#> $fit_type
#> [1] "DRC3pHillFitModelFixS0"
#>