why ggsurvplot can not be called in function frame

[ShixiangWang]

Expected behavior

I wanna call ggsurvplot function instead of default plot function in my custom function

Actual behavior

A error occur.

Steps to reproduce the problem

1. I debug my function and find something wrong when call ggsurvplot.

2. I store my dataset used to call ggsurvplot
   df2.zip

3. I reduce anything else not with this problem and create a minimal function, find the same problem.

test_fun <- function(df2){
    sfit <- survfit(Surv(df2[,"OS"], df2[, "OS_IND"])~group, data=df2)
    ggsurvplot(sfit)
}

test_fun(df2=df2)

> test_fun(df2=df2)
 Show Traceback
 
 Error in is.data.frame(data) : object 'df2' not found 

4. I debug it in function and return

Error in eval(fit$call$data) : object 'df2' not found

But it is very strange that I can use eval(fit$call$data) inside the function.

5. I directly use ggsurvplot in global environment, it works well.

6. I try put df2 in global environment, it works well, and if I change the value of df2， error occur.

I figure it out that ggsurvplot function will retrieve origin data.frame in global environment instead of current frame, can you fix it?

session_info()

R version 3.4.2 (2017-09-28)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

Matrix products: default

locale:
[1] LC_COLLATE=Chinese (Simplified)_People's Republic of China.936 
[2] LC_CTYPE=Chinese (Simplified)_People's Republic of China.936   
[3] LC_MONETARY=Chinese (Simplified)_People's Republic of China.936
[4] LC_NUMERIC=C                                                   
[5] LC_TIME=Chinese (Simplified)_People's Republic of China.936    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] survminer_0.4.2 ggpubr_0.1.6    magrittr_1.5    ggplot2_2.2.1   survival_2.41-3

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.13        compiler_3.4.2      plyr_1.8.4          bindr_0.1           tools_3.4.2        
 [6] digest_0.6.12       tibble_1.3.4        nlme_3.1-131        gtable_0.2.0        lattice_0.20-35    
[11] pkgconfig_2.0.1     rlang_0.1.2         Matrix_1.2-11       psych_1.7.8         cmprsk_2.2-7       
[16] yaml_2.1.14         parallel_3.4.2      bindrcpp_0.2        gridExtra_2.3       knitr_1.17         
[21] dplyr_0.7.4         stringr_1.2.0       survMisc_0.5.4      grid_3.4.2          data.table_1.10.4-1
[26] glue_1.1.1          R6_2.2.2            KMsurv_0.1-5        km.ci_0.5-2         foreign_0.8-69     
[31] reshape2_1.4.2      tidyr_0.7.1         purrr_0.2.3         scales_0.5.0        splines_3.4.2      
[36] assertthat_0.2.0    mnormt_1.5-5        xtable_1.8-2        colorspace_1.3-2    labeling_0.3       
[41] stringi_1.1.5       lazyeval_0.2.0      munsell_0.4.3       broom_0.4.2         zoo_1.8-0   

[kassambara]

Hi, when using ggsurvplot, you should always specify the argument data. In your case, specifying data = df2, should fix the issue

[ShixiangWang]

Soga. thanks. 来自 魅族 PRO 6s

…

-------- 原始邮件 -------- 发件人：Alboukadel KASSAMBARA <notifications@github.com> 时间：周四 2月1日 21:21 收件人：kassambara/survminer <survminer@noreply.github.com> 抄送：王诗翔 <wangshx@shanghaitech.edu.cn>,Author <author@noreply.github.com> 主题：Re: [kassambara/survminer] why ggsurvplot can not be called in function frame (#288) Hi, when using ggsurvplot, you should always specify the argument data. In your case, specifying data = df2, should fix the issue — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#288 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AX5Y5K0FQrO6Ll-tUchhSpl6p67UNmBRks5tQbmqgaJpZM4R1lTp>.

[ShixiangWang]

Hello,
@kassambara it is still something wrong with the result

I used following coding to test [df2.zip]

load("./df2.Rdata")
test_fun <- function(df2){
    sfit <- survfit(Surv(df2[,"OS"], df2[, "OS_IND"])~group, data=df2)
    ggsurvplot(sfit, data=df2)
}
df1 <- df2; rm(df2)
test_fun(df2=df1)

the error is still

object 'df2' not found

I try debug function as deep as I can, and find in .extra.survfit function the formula is still what I parse to function, but the names of data.frame have been changed to data, so the function still can not find df2.

Browse[8]> stats::get_all_vars(.formula, data = data)
Error in eval(inp, data, env) : object 'df2' not found
Browse[8]> .formula
Surv(df2[, "OS"], df2[, "OS_IND"]) ~ df2$group
<environment: 0x000000001ad19d48>

I already know what is wrong,

if I change Surv(df2[,"OS"], df2[, "OS_IND"] to Surv(OS, OS_IND), your function works well.

test_fun <- function(df2){
    sfit <- survfit(Surv(OS, OS_IND)~group, data=df2)
    ggsurvplot(sfit, data=df2)
}

But I cannot do this because I want to write a function which I do not know which variable is mapping to time, which mapping to event in survival analysis.

I tried to use as.symbol function,

test_fun <- function(df2, time="OS", event="OS_IND"){
    sfit <- survfit(Surv(as.symbol(time), as.symbol(event))~group, data=df2)
    ggsurvplot(sfit, data=df2)
}

but

>test_fun(df2=df1)
 Show Traceback
 
 Rerun with Debug
 Error in Surv(as.symbol(time), as.symbol(event)) : 
  Time variable is not numeric 

I use

    colnames(df2)[colnames(df2) == time] <- "time"
    colnames(df2)[colnames(df2) == event] <- "event"
    sfit <- survfit(Surv(time, event) ~ group, data=df2)

to fix this.

It is strange that I cannot use

sfit <- survfit(as.formula(paste0("Surv(", time, ",", event, ") ~ group")), data=df2)

or

.formula <- as.formula(paste0("Surv(", time, ",", event, ") ~ group"))
sfit <- survfit(.formula, data=df2

I know little about advanced R. Can you give me a suggestion to set a right formula that ggsurvplot can rightly work?

[youyuxiansen]

you should declare the "df2" in global env like
$ options(df2='df2')
before you use it in
$ test_fun(df2=df1)

[JanaJarecki]

I second this issue, I don't think it makes sense that a function argument to ggsurvplot has to be declared in the global environment. This seems (IMHO), with all due respect, not perfectly in line with the expected R scoping rules.

[rogercherry]

This stack exchange answer fixed my problem, [how to fix 'Error: variable lengths differ (found for 'input$s')' in R Shiny](https://stackoverflow.com/questions/54058769/how-to-fix-error-variable-lengths-differ-found-for-inputs-in-r-shiny/54225821#54225821)