Estimate effective doses for grouped dose-response data

`estimate_EC50()` fits one dose-response model per isolate, optionally within strata such as year, site, treatment, or fungicide. `ec50_multimodel()` repeats the same workflow for several `drc` model functions and returns model-selection statistics with the estimates.

Usage

estimate_EC50(
  formula,
  data,
  EC_lvl = 50,
  isolate_col,
  strata_col = NULL,
  fct,
  interval = c("none", "delta", "fls", "tfls"),
  type = c("relative", "absolute"),
  quiet = FALSE
)

ec50_multimodel(
  formula,
  data,
  EC_lvl = 50,
  isolate_col,
  strata_col = NULL,
  fct,
  interval = c("none", "delta", "fls", "tfls"),
  type = c("relative", "absolute"),
  quiet = FALSE
)

Arguments

formula

A two-sided formula identifying one numeric response and one numeric dose column, for example `growth ~ dose`.

data

A data frame containing the numeric response, numeric dose, isolate, and optional stratification columns.

EC_lvl

Numeric effective-dose level(s) passed to [drc::ED()]. The default estimates EC50.

isolate_col

Character scalar naming the column that identifies each isolate.

strata_col

Optional character vector naming columns used to split the data before fitting models.

fct

A `drc` model function object such as `drc::LL.3()` for `estimate_EC50()`. For `ec50_multimodel()`, provide a list such as `list(drc::LL.3(), drc::LL.4())`.

interval

Character scalar passed to [drc::ED()]. One of `"none"`, `"delta"`, `"fls"`, or `"tfls"`.

type

Character scalar passed to [drc::ED()]. One of `"relative"` or `"absolute"`.

quiet

Logical. If `FALSE`, failed isolate/model fits are reported with a warning.

Value

A data frame with one row per successful estimate. The first columns identify the isolate (`ID`) and strata, followed by columns returned by [drc::ED()]. `ec50_multimodel()` also appends model-selection statistics from [drc::mselect()] and a `model` column. The result keeps its data-frame behavior, but also stores the original data, formula, grouping columns, model functions, and fitted `drc` models so it can be passed directly to [plot_EC50_curves()].

Examples

data(multi_isolate)
sample_data <- subset(
  multi_isolate,
  isolate == 1 & fungicida == "Fungicide A"
)

estimate_EC50(
  growth ~ dose,
  data = sample_data,
  isolate_col = "isolate",
  strata_col = c("field", "fungicida"),
  fct = drc::LL.3()
)
#>   ID   field   fungicida    Estimate   Std..Error
#> 1  1 Organic Fungicide A 0.006072082 0.0005740341

ec50_multimodel(
  growth ~ dose,
  data = sample_data,
  isolate_col = "isolate",
  strata_col = c("field", "fungicida"),
  fct = list(drc::LL.3(), drc::LL.4())
)
#>    ID   field   fungicida    Estimate   Std..Error    logLik       IC
#> 1   1 Organic Fungicide A 0.006072082 0.0005740341 -45.15079 98.30158
#> 11  1 Organic Fungicide A 0.006364103 0.0007031475 -44.69257 99.38515
#>    Lack.of.fit   Res.var model
#> 1    0.7271292 0.8451681  LL.3
#> 11   0.7370811 0.8498845  LL.4