Targets cannot find data when plotting survival fits using survminer

[arnold-c]

Prework

• Read and agree to the code of conduct and contributing guidelines.
• Confirm that your issue is a genuine bug in the targets package itself and not a user error, known limitation, or issue from another package that targets depends on. For example, if you get errors running tar_make_clustermq(), try isolating the problem in a reproducible example that runs clustermq and not targets. And for miscellaneous troubleshooting, please post to discussions instead of issues.
• If there is already a relevant issue, whether open or closed, comment on the existing thread instead of posting a new issue.
• Post a minimal reproducible example like this one so the maintainer can troubleshoot the problems you identify. A reproducible example is:
  • Runnable: post enough R code and data so any onlooker can create the error on their own computer.
  • Minimal: reduce runtime wherever possible and remove complicated details that are irrelevant to the issue at hand.
  • Readable: format your code according to the tidyverse style guide.

Description

When using the {survminer} package to plot survival curves with the ggsurvplot()function, {targets} cannot find the data if it is a data.frame/tibble created within a notebook (specified in _targets.R with tar_render()). This is not an issue when the notebook is knit normally. There is a workaround by using a modified plotting function that takes a summary data.frame (see example 3 below). I don't believe this is expected behavior with regards to the way {targets} accesses data.frames in memory, but please close if it is.

Reproducible example

library(survival)
library(survminer)
attach(lung)

# Create a data.frame in the notebook
test <- filter(lung, sex == 1)

# This works in a targets pipeline
reprex_survival_1 <- survfit(Surv(time, status) ~ 1, data = lung)
ggsurvplot(reprex_survival_1)

# This doesn't work in a targets pipeline
reprex_survival_2 <- survfit(Surv(time, status) ~ 1, data = test)
ggsurvplot(reprex_survival_2)

# This also works in a targets pipeline - presumably because the `surv_summary()`
# function creates a data.frame
reprex_survival_3 <- survfit(Surv(time, status) ~ 1, data = test)
ggsurvplot_df(surv_summary(reprex_survival_1))

Expected result

It should create the survival plot, as it does with examples 1 and 3 above.

Diagnostic information

sessioninfo::session_info()
#> ─ Session info ───────────────────────────────────────────────────────────────
#>  setting  value                       
#>  version  R version 4.1.0 (2021-05-18)
#>  os       macOS Big Sur 10.16         
#>  system   x86_64, darwin17.0          
#>  ui       X11                         
#>  language (EN)                        
#>  collate  en_US.UTF-8                 
#>  ctype    en_US.UTF-8                 
#>  tz       America/Toronto             
#>  date     2021-07-01                  
#> 
#> ─ Packages ───────────────────────────────────────────────────────────────────
#>  package     * version date       lib source             
#>  cli           2.5.0   2021-04-26 [1] standard (@2.5.0)  
#>  digest        0.6.27  2020-10-24 [1] standard (@0.6.27) 
#>  evaluate      0.14    2019-05-28 [1] standard (@0.14)   
#>  fs            1.5.0   2020-07-31 [1] standard (@1.5.0)  
#>  glue          1.4.2   2020-08-27 [1] standard (@1.4.2)  
#>  highr         0.9     2021-04-16 [1] standard (@0.9)    
#>  htmltools     0.5.1.1 2021-01-22 [1] standard (@0.5.1.1)
#>  knitr         1.33    2021-04-24 [1] standard (@1.33)   
#>  magrittr      2.0.1   2020-11-17 [1] standard (@2.0.1)  
#>  reprex        2.0.0   2021-04-02 [1] standard (@2.0.0)  
#>  rlang         0.4.11  2021-04-30 [1] standard (@0.4.11) 
#>  rmarkdown     2.8     2021-05-07 [1] standard (@2.8)    
#>  rstudioapi    0.13    2020-11-12 [1] standard (@0.13)   
#>  sessioninfo   1.1.1   2018-11-05 [1] standard (@1.1.1)  
#>  stringi       1.6.2   2021-05-17 [1] standard (@1.6.2)  
#>  stringr       1.4.0   2019-02-10 [1] standard (@1.4.0)  
#>  withr         2.4.2   2021-04-18 [1] standard (@2.4.2)  
#>  xfun          0.23    2021-05-15 [1] standard (@0.23)   
#>  yaml          2.2.1   2020-02-01 [1] standard (@2.2.1)  
#> 
#> [1] /Library/Frameworks/R.framework/Versions/4.1/Resources/library

> traceback()
8: stop(error_run(...))
7: throw_run(conditionMessage(e), "\nVisit https://books.ropensci.org/targets/debugging.html ", 
       "for debugging advice.")
6: value[[3L]](cond)
5: tryCatchOne(expr, names, parentenv, handlers[[1L]])
4: tryCatchList(expr, classes, parentenv, handlers)
3: tryCatch(callr_dispatch(targets_function, targets_arguments, 
       callr_function, callr_arguments), callr_error = function(e) {
       throw_run(conditionMessage(e), "\nVisit https://books.ropensci.org/targets/debugging.html ", 
           "for debugging advice.")
   })
2: callr_outer(targets_function = tar_make_inner, targets_arguments = targets_arguments, 
       callr_function = callr_function, callr_arguments = callr_arguments)
1: (function (names = NULL, reporter = Sys.getenv("TAR_MAKE_REPORTER", 
       unset = "verbose"), callr_function = callr::r, callr_arguments = targets::callr_args_default(callr_function, 
       reporter)) 
   {
       assert_script()
       assert_flag(reporter, tar_make_reporters())
       assert_callr_function(callr_function)
       assert_list(callr_arguments, "callr_arguments mut be a list.")
       targets_arguments <- list(names_quosure = rlang::enquo(names), 
           reporter = reporter)
       out <- callr_outer(targets_function = tar_make_inner, targets_arguments = targets_arguments, 
           callr_function = callr_function, callr_arguments = callr_arguments)
       invisible(out)
   })()

15f6baa06e6a043d7aa930f8ba23778e2d740162"

[wlandau]

I assume you mean a pipeline like this?

library(targets)
tar_script({
  library(survival)
  library(survminer)
  library(targets)
  list(
    tar_target(test, lung),
    tar_target(survival, survfit(Surv(time, status) ~ 1, data = test)),
    tar_target(plot, ggsurvplot(survival))
  )
})

tar_make()
#> Loading required package: ggplot2
#> Loading required package: ggpubr
#> • start target test
#> • built target test
#> • start target survival
#> • built target survival
#> • start target plot
#> x error target plot
#> • end pipeline
#> Error : object 'test' not found
#> Error: callr subprocess failed: object 'test' not found
#> Visit https://books.ropensci.org/targets/debugging.html for debugging advice.

^{Created on 2021-07-01 by the reprex package (v2.0.0)}

This is an instance of #160 and not a bug in targets. For reproducibility purposes, targets runs each target in a non-global environment that inherits from globalenv(). Some packages like lme4 and apparently survminer/survival expect data to be in globalenv() despite claims in the docs that data = NULL means the data is taken from the fit object. I can reproduce this error without targets, so I recommend following up with an issue in the survminer or survival.

library(survival)
library(survminer)
#> Loading required package: ggplot2
#> Loading required package: ggpubr
envir <- new.env(parent = globalenv())
evalq({
  test <- lung
  survival <- survfit(Surv(time, status) ~ 1, data = test)
  ggsurvplot(survival)
}, envir = envir)
#> Error in eval(fit$call$data): object 'test' not found

^{Created on 2021-07-01 by the reprex package (v2.0.0)}

ggsurvplot() takes a data argument, so I recommend passing the data explicitly. The pipeline works if I do that.

library(targets)
tar_script({
  library(survival)
  library(survminer)
  library(targets)
  list(
    tar_target(test, lung),
    tar_target(survival, survfit(Surv(time, status) ~ 1, data = test)),
    tar_target(plot, ggsurvplot(survival, data = test))
  )
})

tar_make()
#> Loading required package: ggplot2
#> Loading required package: ggpubr
#> • start target test
#> • built target test
#> • start target survival
#> • built target survival
#> • start target plot
#> • built target plot
#> • end pipeline

^{Created on 2021-07-01 by the reprex package (v2.0.0)}