imap-ing over multiple glmers in R: Anonymous functions

Question

I have this function:

mweFitModelsGLMER <- function(longData,
 strModelName = "ID Rand Model",
 strFormula = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 ...)
{
 output = list() # Initialise output
 Ptn <- unique(longData$Ptn)
 
 Model <- longData %>%
 lme4::glmer(formula = stats::as.formula(strFormula),
 ...)
 #Do a few other things like prepare summary table etc.
 
 output$Models <-
 # list(
 tibble(
 Protein = Ptn,
 Formula = strFormula,
 `Model Name` = strModelName,
 Model = list(Model),
 )
 #)
 return(output)
}

Whose primary purpose is to fit a mixed model, do a few other things and then output a big tibble. This works well on its own with the data at the end of the post:

mweFitModelsGLMER(allData, family = "binomial", na.action = na.pass)

Now I try to create a helper function, which can take in a named list of formula specifications, pass it on to the function above and return the output:

mweMultipleModelsGLMER <- function(longData,
 namLstFormula = list(Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"),
 ...)
{
 purrr::imap(
 namLstFormula,
 ~ mweFitModelsGLMER(
 longData = longData,
 strModelName = paste(.y),
 strFormula = .x,
 ...
 )
 )
}

And run it like so:

mweMultipleModelsGLMER(allData, 
 namLstFormula = 
 list(Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"), 
 family = "binomial")

To which, R responds with

Error in map2(): i In index: 1. i With name: Model1. Caused by error in lme4::glmer(): ! 'control' is not a list; use glmerControl()

Ok, then I be explicit in my call..

mweMultipleModelsGLMER(
 allData,
 namLstFormula = list(Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"),
 family = "binomial",
 control = lme4::glmerControl(optimizer = "bobyqa")
)

To which R comes back to me with:

Error in map2(): i In index: 1. i With name: Model1. Caused by error in lme4::glmer(): ! 'control' is not a list; use glmerControl()

Backtrace:
 ▆
 1. ├─global mweMultipleModelsGLMER(...)
 2. │ └─purrr::imap(...)
 3. │ └─purrr::map2(.x, vec_index(.x), .f, ...)
 4. │ └─purrr:::map2_("list", .x, .y, .f, ..., .progress = .progress)
 5. │ ├─purrr:::with_indexed_errors(...)
 6. │ │ └─base::withCallingHandlers(...)
 7. │ ├─purrr:::call_with_cleanup(...)
 8. │ └─.f(.x[[i]], .y[[i]], ...)
 9. │ └─global mweFitModelsGLMER(...)
 10. │ └─longData %>% ...
 11. └─lme4::glmer(., formula = stats::as.formula(strFormula), ...)
 12. └─base::stop("'control' is not a list; use glmerControl()")

For reference, I did help("glmer")

glmer(formula, data = NULL, family = gaussian , control = glmerControl() , start = NULL , verbose = 0L , nAGQ = 1L , subset, weights, na.action, offset, contrasts = NULL , mustart, etastart , devFunOnly = FALSE)

What am I missing?

Data below:

library(tidyverse)
allData <- structure(list(phMeta_UID = c("Pat 1 BRS", "Pat 1 BRS", "Pat 1 BRS", 
"Pat 1 BRS", "Pat 1 BRS", "Pat 1 BRS", "Pat 1 BRS", "Pat 1 BRS", 
"Pat 1 BRS", "Pat 1 SHF", "Pat 1 SHF", "Pat 1 SHF", "Pat 1 SHF", 
"Pat 1 SHF", "Pat 1 SHF", "Pat 1 SHF", "Pat 2 BRS", "Pat 2 BRS", 
"Pat 2 BRS", "Pat 2 SHF", "Pat 2 SHF", "Pat 2 SHF", "Pat 2 SHF", 
"Pat 2 SHF", "Pat 2 SHF", "Pat 3 SHF", "Pat 3 SHF", "Pat 3 SHF", 
"Pat 3 SHF", "Pat 3 SHF", "Pat 3 SHF", "Pat 3 SHF", "Pat 3 SHF", 
"Pat 1 SHF", "Pat 2 BRS", "Pat 2 BRS", "Pat 2 BRS", "Pat 2 BRS", 
"Pat 2 BRS", "Pat 2 BRS", "Pat 2 SHF"), phMeta_Time = c(0, 0.5, 
1, 2, 3, 4, 6, 8, 12, 0, 0.5, 2, 3, 4, 8, 12, 0, 0.5, 2, 0, 0.5, 
2, 3, 4, 10, 3, 4, 8, 9, 12, 0, 0.5, 2, 1, 1, 3, 4, 6, 8, 12, 
1), phMeta_Batch = c(1, 1, 2, 1, 1, 1, 2, 2, 2, 2, 2, 1, 2, 1, 
1, 2, 1, 2, 2, 1, 1, 2, 2, 2, 1, 2, 1, 1, 1, 1, 1, 1, 2, 1, 1, 
1, 2, 1, 1, 2, 2), phMeta_Site = c("BRS", "BRS", "BRS", "BRS", 
"BRS", "BRS", "BRS", "BRS", "BRS", "SHF", "SHF", "SHF", "SHF", 
"SHF", "SHF", "SHF", "BRS", "BRS", "BRS", "SHF", "SHF", "SHF", 
"SHF", "SHF", "SHF", "SHF", "SHF", "SHF", "SHF", "SHF", "SHF", 
"SHF", "SHF", "SHF", "BRS", "BRS", "BRS", "BRS", "BRS", "BRS", 
"SHF"), phMeta_Toxicity = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L), levels = c("N", "Y"), class = "factor"), phMeta_Patient = c(4, 
4, 4, 4, 4, 4, 4, 4, 4, 1, 1, 1, 1, 1, 1, 1, 5, 5, 5, 2, 2, 2, 
2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 1, 5, 5, 5, 5, 5, 5, 2), phMeta_SiteXPatient = c(1, 
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 1, 2, 2, 2, 2, 2, 2, 2), phMeta_Phlebotomy = c(NA, 
NA, NA, NA, NA, NA, NA, NA, NA, "Venous", "Capillary", "Venous", 
NA, "Venous", "Capillary", "Venous", NA, NA, NA, "Venous", "Capillary", 
"Capillary", "Capillary", "Capillary", "Venous", "Venous", "Venous", 
"Venous", NA, "Venous", "Venous", "Capillary", "Venous", "Venous", 
NA, NA, NA, NA, NA, NA, "Capillary"), value = structure(c(0.821944197243299, 
-0.0198715543825022, 0.631293040769747, 0.0849008002934989, 0.0887812578740912, 
1.05453893286552, -2.12824809442977, 0.276304235154362, 0.859985670512456, 
-1.04724028808727, -0.00411159277206202, 1.67249503748148, -0.54509175851945, 
-1.46818604842327, -0.498407201908304, 0.51962029081445, -1.08569307903582, 
-0.270959349353233, -0.211864905388811, 0.158115347225517, 0.0227322978830837, 
0.852349233070034, -0.401612245382643, -2.98257656282869, -0.191542564781942, 
-1.45732401444245, 0.494605681417659, 0.464925863604591, 0.856056504259303, 
1.63695935481179, -1.33574175565861, 0.42694933523218, -0.0213328145592944, 
-0.966697791972374, -0.634006734239892, 0.128202810199108, 0.960861331383678, 
-0.258051551124902, 1.25488311517846, 1.08015428255721, 1.18190128745978
), dim = c(41L, 1L)), Ptn = c("phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand", 
"phOSI_ICOS ligand", "phOSI_ICOS ligand", "phOSI_ICOS ligand"
)), row.names = c(NA, -41L), class = c("tbl_df", "tbl", "data.frame"
))

EDIT 1:

The function works if I ignore the list names:

fFitMultipleModelsGLMER2 <- function(nestlistData,
 strPredictorROC = "phMeta_Toxicity",
 LstFormula = c("phMeta_Toxicity ~ 1 + (1 | phMeta_UID)", 
 "phMeta_Toxicity ~ value + (1 | phMeta_UID)"),
 ...)
{
 purrr::map(
 namLstFormula,
 ~fFitModelsGLMER(
 longData = nestlistData,
 strPredictor = strPredictorROC,
 strFormula = .x,
 ...
 )
 )
}

EDIT2: Editing to improve discoverability

Answer 1

The issue is the use of a purrr-style anonymous function, i.e. using a formula, and .... As an unintended side-effect, instead of the optional arguments you passed via ... to the "outer" function, you are passing the .x and .y arguments to the "inner" mweFitModelsGLMER a second time via ....

This can be seen from the following example where I simply print the content of ... inside the imap call:

library(purrr)
foo <- function(longData,
 namLstFormula = list(
 Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"
 ),
 ...) {
 purrr::imap(
 namLstFormula,
 ~ print(list(...))
 )
}
baz <- foo(allData,
 namLstFormula =
 list(
 Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"
 ),
 family = "binomial"
)
#> [[1]]
#> [1] "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)"
#> 
#> [[2]]
#> [1] "Model1"
#> 
#> [[1]]
#> [1] "phMeta_Toxicity ~ value + (1 | phMeta_UID)"
#> 
#> [[2]]
#> [1] "Model2"

To fix your issue use function(.x, .y) or \(.x, .y) instead of a formula:

mweMultipleModelsGLMER <- function(longData,
 namLstFormula = list(
 Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"
 ),
 ...) {
 purrr::imap(
 namLstFormula,
 \(.x, .y)
 mweFitModelsGLMER(
 longData = longData,
 strModelName = .y,
 strFormula = .x,
 ...
 )
 )
}
mweMultipleModelsGLMER(allData,
 namLstFormula =
 list(
 Model1 = "phMeta_Toxicity ~ 1 + (1 | phMeta_UID)",
 Model2 = "phMeta_Toxicity ~ value + (1 | phMeta_UID)"
 ),
 family = "binomial"
)
#> $Model1
#> $Model1$Models
#> # A tibble: 1 ×ばつ 4
#> Protein Formula `Model Name` Model 
#> <chr> <chr> <chr> <list> 
#> 1 phOSI_ICOS ligand phMeta_Toxicity ~ 1 + (1 | phMeta_U... Model1 <glmerMod>
#> 
#> 
#> $Model2
#> $Model2$Models
#> # A tibble: 1 ×ばつ 4
#> Protein Formula `Model Name` Model 
#> <chr> <chr> <chr> <list> 
#> 1 phOSI_ICOS ligand phMeta_Toxicity ~ value + (1 | phMe... Model2 <glmerMod>