[Bioc-devel] ddply causes error during R check

Hi,

I am developing a Bioconductor package and can not get rid of some
warning messages. During devtools::check() I get the following warning
messages:

...
summarizeDataFrame: no visible binding for global variable ?name?
summarizeDataFrame: no visible binding for global variable ?gene?
summarizeDataFrame: no visible binding for global variable ?value?
...

Here a short version of the function:

#' Collapse rows with duplicated name column
#'
#' @param dat a \cite{tibble} with the columns name, gene and value
#' @importFrom plyr ddply
#' @import tibble
#' @return a \cite{tibble}
#' @export
#'
#' @examples
#' dat <- tibble(name = c(paste0("position", 1:5), paste0("position",
c(1:3))), gene = paste0("gene", 1:8), value = 1:8)
#' summarizeDataFrame(dat)
summarizeDataFrame <- function(dat){
   ddply(dat, "name", "summarize",
         name=unique(name),
         gene=paste(unique(gene), collapse = ","),
         value=mean(value))
}

R interprets the "name", "gene" and "value" column names as variables
during the check. Does anyone has an idea how to change the syntax of
ddply or how to get rid of the warning message?

Thanks in advance!

Tobias

hpx = 1
ddply(mtcars, "cyl", "summarize", value = mean(hpx)).  ## oops, meant `mean(hp)`.

aggregate(hp ~ cyl, mtcars, mean)

If you're sure these are false positives (and it looks like they are) 
then you can use utils::globalVariables() outside of your function to 
get rid of the note.? It might also be worth pointing out that there 
are also plenty of Bioconductor packages that don't do this and simply 
have this mentioned in the check results e.g 
http://bioconductor.org/checkResults/devel/bioc-LATEST/beadarray/malbec2-checksrc.html 
.

Mike

On Tue, 12 Feb 2019 at 08:35, web working <webworking at posteo.de 
<mailto:webworking at posteo.de>> wrote:

    Hi,

    I am developing a Bioconductor package and can not get rid of some
    warning messages. During devtools::check() I get the following
    warning
    messages:

    ...
    summarizeDataFrame: no visible binding for global variable ?name?
    summarizeDataFrame: no visible binding for global variable ?gene?
    summarizeDataFrame: no visible binding for global variable ?value?
    ...

    Here a short version of the function:

    #' Collapse rows with duplicated name column
    #'
    #' @param dat a \cite{tibble} with the columns name, gene and value
    #' @importFrom plyr ddply
    #' @import tibble
    #' @return a \cite{tibble}
    #' @export
    #'
    #' @examples
    #' dat <- tibble(name = c(paste0("position", 1:5), paste0("position",
    c(1:3))), gene = paste0("gene", 1:8), value = 1:8)
    #' summarizeDataFrame(dat)
    summarizeDataFrame <- function(dat){
    ?? ddply(dat, "name", "summarize",
    ???????? name=unique(name),
    ???????? gene=paste(unique(gene), collapse = ","),
    ???????? value=mean(value))
    }

    R interprets the "name", "gene" and "value" column names as variables
    during the check. Does anyone has an idea how to change the syntax of
    ddply or how to get rid of the warning message?

    Thanks in advance!

    Tobias

use `globalVariables()` to declare these symbols and quieten warnings, at the expense of quietening warnings about undefined variables in _all_ code and potentially silencing true positives. Avoid non-standard evaluation (this is what ddply is doing, using special rules to resolve symbols like `name`) by using base R functionality; note also that non-standard evaluation is prone to typos, e.g., looking for the typo `hpx` in the calling environment rather than the data frame

hpx = 1
ddply(mtcars, "cyl", "summarize", value = mean(hpx)).  ## oops, meant `mean(hp)`.

   cyl summarize
1   4         1
2   6         1
3   8         1

Marginally better is

aggregate(hp ~ cyl, mtcars, mean)

   cyl        hp
1   4  82.63636
2   6 122.28571
3   8 209.21429

where R recognizes symbols in the formula ~ as intentionally unresolved. The wizards on the list might point to constructs in the rlang package.

Martin

?On 2/12/19, 2:35 AM, "Bioc-devel on behalf of web working" <bioc-devel-bounces at r-project.org on behalf of webworking at posteo.de> wrote:

     Hi,
     
     I am developing a Bioconductor package and can not get rid of some
     warning messages. During devtools::check() I get the following warning
     messages:
     
     ...
     summarizeDataFrame: no visible binding for global variable ?name?
     summarizeDataFrame: no visible binding for global variable ?gene?
     summarizeDataFrame: no visible binding for global variable ?value?
     ...
     
     Here a short version of the function:
     
     #' Collapse rows with duplicated name column
     #'
     #' @param dat a \cite{tibble} with the columns name, gene and value
     #' @importFrom plyr ddply
     #' @import tibble
     #' @return a \cite{tibble}
     #' @export
     #'
     #' @examples
     #' dat <- tibble(name = c(paste0("position", 1:5), paste0("position",
     c(1:3))), gene = paste0("gene", 1:8), value = 1:8)
     #' summarizeDataFrame(dat)
     summarizeDataFrame <- function(dat){
        ddply(dat, "name", "summarize",
              name=unique(name),
              gene=paste(unique(gene), collapse = ","),
              value=mean(value))
     }
     
     R interprets the "name", "gene" and "value" column names as variables
     during the check. Does anyone has an idea how to change the syntax of
     ddply or how to get rid of the warning message?
     
     Thanks in advance!
     
     Tobias