[Feature Request]: Add facet + cohort legend strata order in mean plot

[legrasv]

Feature description

Dear NEST,

For a project, we needed to generate a mean plot by Nominal dose (group), but facet by COUNTRY. There was no way of adding a facet to the plot aftewards as the data in the plot is the summarized data with the mean and sd, and the column COUNTRY is lost. We had to manually modify the function so that a second grouping column is kept:

These are the code we added to g_lineplot:

`#'   * `facetvar` (`character`)\cr name of grouping variable, i.e. treatment arm. Can be `NA` to indicate lack of groups.



  if (is.na(variables["facetvar"])) {
    facetvar <- NULL # NULL if facetvar == NA or it is not in variables
  } else {
    facetvar <- variables[["facetvar"]]
  }




  ####################################### |
  # ---- Compute required statistics ----
  ####################################### |
  if (!is.null(facetvar) & !is.null(strata)) {
    df_grp <- tidyr::expand(df, .data[[facetvar]], .data[[strata]], .data[[x]]) # expand based on levels of factors
  } else  if (!is.null(strata)) {
    df_grp <- tidyr::expand(df, NULL, .data[[strata]], .data[[x]]) # expand based on levels of factors
  } else {
    df_grp <- tidyr::expand(df, NULL, NULL, .data[[x]])
  }
  df_grp <- df_grp %>%
    dplyr::full_join(y = df[, c(facetvar, strata, x, y)], by = c(facetvar, strata, x), multiple = "all") %>%
    dplyr::group_by_at(c(facetvar, strata, x))



#' @export
control_lineplot_vars <- function(x = "AVISIT", y = "AVAL", strata = "ARM", paramcd = "PARAMCD", y_unit = "AVALU",
                                  cohort_id = "USUBJID", facetvar=NA) {
  checkmate::assert_string(x)
  checkmate::assert_string(y)
  checkmate::assert_string(strata, na.ok = TRUE)
  checkmate::assert_string(cohort_id, na.ok = TRUE)
  checkmate::assert_string(paramcd, na.ok = TRUE)
  checkmate::assert_string(y_unit, na.ok = TRUE)
  checkmate::assert_string(facetvar, na.ok = TRUE)
  
  variables <- c(x = x, y = y, strata = strata, paramcd = paramcd, y_unit = y_unit, cohort_id = cohort_id, facetvar=facetvar)
  return(variables)
}

And then with the parameter facetvar in the plot, we could produce our plot.

Would it be possible to code something similar to allow us to produce facets on this plot? please

also the legend was incorrect when using strata:

Could you check this?
issue raised a month ago:
https://insightsengineering.github.io/tlg-catalog/stable/graphs/pharmacokinetic/pkcg03.html

Best regards
Valentin

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.

[ayogasekaram]

Hi @legrasv, I have a PR going up for this soon so I can tag you for review. I just wanted to confirm if you have any warning message in mind in the case of the alt_counts_df? for example, if the levels for the facet_var are set in adpc for example, but the user also supplied an alt_counts_df, the same levels need to be set there for the plot legend to reflect it.

[legrasv]

@ayogasekaram Thank you for working on that, a colleague will test the new function shortly.
For your question, I believe that alt_counts_df is always equal to df (in our case), I don't see any case when you would like to put two different datasets in the two arguments. So I think we are fine from this side, I don't think we need a warning message.

[legrasv]

@ayogasekaram I got this error message which I can't debug, it looks like I don't have the function sfun loaded (when I am using browser()), what is this function? how can I get it?

[ayogasekaram]

@legrasv the sfun is s_summary but that default should already be set in the g_lineplot parameters. Let me take a look on my end as well!

[legrasv]

Dear NEST,

I found a bug in the code of g_lineplot for the strata:

matches <- sapply(unique(df_N[[group_var]]), function(x) unique(df_N[[paste0(group_var, "_N")]])[grepl(paste0("^", x), unique(df_N[[paste0(group_var, "_N")]]))])

in this piece of code, it happens that group_var contains specific regex characters and therefore the grepl will not work, example:

Browse[2]> grepl("^PART 1 \\(Period Effect\\)", unique(df_N[[paste0(group_var, "_N")]]))
[1]  TRUE  TRUE FALSE FALSE
Browse[2]> grepl("^PART 1 (Period Effect)", unique(df_N[[paste0(group_var, "_N")]]))
[1] FALSE FALSE FALSE FALSE

-> I need to tell that "()" are not part of the regex expression. I don't know how we could handle that in general. Here I have the bug with the "(", but it could happen with regex characters such as "+", ".", "[", etc.

[edelarua]

@ayogasekaram can this issue be closed now?

[ayogasekaram]

yes it's been addressed in this PR