Eta_Cats

[biethbr1]

cats.labels (+use.labels) does not work.
How to change covariate labels to full words?
How to specify symmetric ranges of y-axis different for each parameter?
Does not work to add hline.
Shrinkage is not shown.
Jitter does not work.

(Old Ticket #235)

[KZRKZR]

Hi @biethbr1 Please check findings/questions/solutions:

1. cats.labels (+use.labels) does not work. Unfortunately in the current implementation, there were no possibilities to set strat.facet for these types of plots: eta_matrix, eta_cats, eta_conts. Not sure could we use cats.labels without strat.facet - will discover more.
2. How to change covariate labels to full words? We could do it in different ways:

2.1 Workaround approach: by adding facet_grid function with labeller argument.

mlxpath <- file.path(system.file(package = "ggPMX"),
                     "testdata", "1_popPK_model", "project.mlxtran")
ctr <- pmx_mlxtran(mlxpath, config = "standing", settings = mySet)
ctr %>% pmx_plot_eta_cats(labels = list(title = "EBE vs discrete covariates")) +
  facet_grid(EFFECT~variable,
             labeller = labeller(
             variable=c('SEX'="Gender", 'RACE'="Race", 'DISE' = "Disease", 'ILOW' = "I low"),
             EFFECT=c('CL'= "first", 'F'= "second", 'Vc'= "third", 'Vp1'= "forth")
             ))

2.2 Make changes in existing code:

• set strat = TRUE in eta_cov structure.
• add code to plot_pmx.eta_cov function:

if (!is.null(x$facets)) 
          p <- p + facet_grid(stats::as.formula("EFFECT~variable"), scales = "free", labeller = x$facets$labeller)

Call will look like:

mySet=pmx_settings(is.draft=FALSE, use.abbrev = FALSE, use.labels = TRUE, 
                                      cats.labels = list(
                                        RACE = c("0"= "As", "1"="Bl"),
                                          variable=c('SEX'="Gender", 'RACE'="Race", 'DISE' = "Disease", 'ILOW' = "I low"),
                                          EFFECT=c('CL'= "first", 'F'= "second", 'Vc'= "third", 'Vp1'= "forth")
                                      )
                   )
mlxpath <- file.path(system.file(package = "ggPMX"),
                     "testdata","1_popPK_model","project.mlxtran")
ctr <- pmx_mlxtran(mlxpath, config = "standing", settings = mySet)

ctr %>% pmx_plot_eta_cats(labels = list(title = "EBE vs discrete covariates"))

Result:

2.3 I also see some comments in the code plot-eta-cov.R:

' In case of pmx_plot_eta_cats and pmx_plot_eta_conts you can customize the covariates

' and covaraites labels using pmx_cov.

Maybe someone could clarify this in details.

[KZRKZR]

3. How to specify symmetric ranges of y-axis different for each parameter? Could you please explain in detail how the result must looks like? Something like this:

[KZRKZR]

4. Does not work to add hline. Will add changes to plot-eta-cov.R Call will be ctr %>% pmx_plot_eta_cats(labels = list(title = "EBE vs discrete covariates"), is.hline=TRUE, hline = 0.25)

[baltcir1]

About 3. Yes, this looks good!

[KZRKZR]

About 3. At first glance it possible to do with workaround:

library(facetscales)
library(dplyr)
library(ggplot2)
library(reshape)

ctr <- theophylline()
df <- ctr$data$eta
df <- df[, c(ctr$cats, "VALUE", "EFFECT"), with = FALSE]
max_y <- df %>%
  group_by(EFFECT) %>%
  summarise(
    maxValue = max(abs(VALUE), na.rm = T)
  ) %>%
  arrange(EFFECT)
get_scales_y <- function(i, scales_y = NULL){
  scales_i <- list(scale_y_continuous(limits = c(-max_y$maxValue[i], max_y$maxValue[i])))
  names(scales_i) <- max_y$EFFECT[i]
  append(scales_y, scales_i)
}
scales_y <- sapply(1:length(max_y$EFFECT), get_scales_y)
p <- ggplot(melt(df, measure.vars = ctr$cats)) +
  geom_boxplot(aes_string(x = "value", y = "VALUE"))
p + facet_grid_sc(stats::as.formula("EFFECT~variable"), scales = list(y = scales_y))

Will check how to implement this in ggPMX. Do we plan to use this functionality in other charts? If yes, which exactly?

[KZRKZR]

@baltcir1 Hi! Are we good to introduce new dependency for implementation of symmetric ranges Y-axis - facetscales library (it is not on CRAN)?

[KZRKZR]

Approach 2.3 seems like is the right way to manage labels. This approach currently works for pmx_plot_eta_conts

mlxpath <- file.path(system.file(package = "ggPMX"),
                     "testdata","1_popPK_model","project.mlxtran")
ctr <- pmx_mlxtran(mlxpath, config = "standing")
ctr %>% pmx_plot_eta_conts(
  covariates=pmx_cov(values=list("HT0","TRT"),
                     labels=list("HT0 long","TRT long"))
)

After improvments will work for pmx_plot_eta_cats

mlxpath <- file.path(system.file(package = "ggPMX"),
                     "testdata","1_popPK_model","project.mlxtran")
ctr <- pmx_mlxtran(mlxpath, config = "standing")
ctr %>% pmx_plot_eta_cats(
  covariates=pmx_cov(values=list("RACE","DISE"),
                     labels=list("Race changed","Dise changed"))
)

[baltcir1]

Dear Kostia, This library is not on CRAN and is not planning to go on CRAN, so it’s not an option for us. Is there another way to solve this? Many thanks, Irina

From: Kostiantyn Zinher notifications@github.com Sent: Thursday, 19 November 2020 10:04 To: ggPMXdevelopment/ggPMX ggPMX@noreply.github.com Cc: Baltcheva, Irina irina.baltcheva@novartis.com; Mention mention@noreply.github.com Subject: Re: [ggPMXdevelopment/ggPMX] Eta_Cats (#71)

@baltcir1https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_baltcir1&d=DwMCaQ&c=ZbgFmJjg4pdtrnL2HUJUDw&r=aKykINevLIXBb2-gHmhroKGVY88hh9HCgL0YAgz8yzg&m=6MUeYFKgsdp2amfiX4CkqT2gXhnHH6iP0N0dTy9r5Ow&s=49KuJw29J8KJvOBGaBOJmLlFpiwyZ7yPStS0pwER1VU&e= Hi! Are we good to introduce new dependency for implementation of symmetric ranges Y-axis - facetscales library?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_ggPMXdevelopment_ggPMX_issues_71-23issuecomment-2D730230837&d=DwMCaQ&c=ZbgFmJjg4pdtrnL2HUJUDw&r=aKykINevLIXBb2-gHmhroKGVY88hh9HCgL0YAgz8yzg&m=6MUeYFKgsdp2amfiX4CkqT2gXhnHH6iP0N0dTy9r5Ow&s=qM5hEXlBDRsDPR-n8uDI2fuNqj9zmIweqa9TkaSyT0Q&e=, or unsubscribehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AGCGRLCMSKFWJECF4TKMSRTSQTNRFANCNFSM4NADGESA&d=DwMCaQ&c=ZbgFmJjg4pdtrnL2HUJUDw&r=aKykINevLIXBb2-gHmhroKGVY88hh9HCgL0YAgz8yzg&m=6MUeYFKgsdp2amfiX4CkqT2gXhnHH6iP0N0dTy9r5Ow&s=3js0LBumGA_GdHnSGOUwcGajSkoyF1dnWJWvD-EGPfw&e=.