add option for generating guest function outputs for residualVsObserv… #176
Conversation
…ed ddi ratio plot
Codecov Report
@@ Coverage Diff @@
## develop #176 +/- ##
===========================================
- Coverage 48.41% 48.24% -0.18%
===========================================
Files 55 55
Lines 2551 2560 +9
===========================================
Hits 1235 1235
- Misses 1316 1325 +9
Continue to review full report at Codecov.
|
…istency with diagonal case
R/obs-vs-pred-datamapping.R
Outdated
| # Apply log10 transformation to lines because | ||
| # plot is log scaled in by default and geom_abline | ||
| # requires the log transformed values in input of intercept | ||
| self$lines <- lapply(lines, log10) |
There was a problem hiding this comment.
Does default value in DefaultDataMappingValues include the log10 ?
If so, line 27 may apply twice the log.
This means that either we need lapply(lines, log10) %||% DefaultDataMappingValues$obsVsPred with lines = NULL in input or that DefaultDataMappingValues account for scale somehow
There was a problem hiding this comment.
I think we may want to apply the log10 directly within the plotXX functions after checking the plotConfiguration scale is included in Scaling$log (obs vs pred and res vs pred might be in linear scale).
Otherwise, users can get unwanted log transformation of their diagonal/horizontal lines.
There was a problem hiding this comment.
If possible could you document the scaling in the @param lines and maybe in the pk-ratio vignette
There was a problem hiding this comment.
Reverting changes to obs-vs-pred-datamapping.R and pkratio-datamapping.R for now.
R/plot-ddiratio.R
Outdated
| if(dataMapping$residualsVsObserved){ | ||
| lineOrientation <- "horizontal" | ||
| } |
There was a problem hiding this comment.
Could you add a shortcut option for users in the function input as well: e.g. residuals = NULL ?
and internally do something like
validateIsLogical(residuals, nullAllowed = TRUE)
residuals <- residuals %||% dataMapping$residualsVsObserved`
lineOrientation <- "diagonal"
if(residuals){lineOrientation <- "horizontal"}
I think such options are quite convenient for plotTornado and plotBoxWhisker and might be valuable for users in plotDDIRatio
There was a problem hiding this comment.
Maybe use residualsVsObserved instead of residuals to keep it consistent with dataMapping
There was a problem hiding this comment.
Done, but with residualsVsObserved = FALSE by default and nullAllowed = FALSE
R/obs-vs-pred-datamapping.R
Outdated
| # Apply log10 transformation to lines because | ||
| # plot is log scaled in by default and geom_abline | ||
| # requires the log transformed values in input of intercept | ||
| self$lines <- lapply(lines, log10) |
There was a problem hiding this comment.
I think we may want to apply the log10 directly within the plotXX functions after checking the plotConfiguration scale is included in Scaling$log (obs vs pred and res vs pred might be in linear scale).
Otherwise, users can get unwanted log transformation of their diagonal/horizontal lines.
R/obs-vs-pred-datamapping.R
Outdated
| # Apply log10 transformation to lines because | ||
| # plot is log scaled in by default and geom_abline | ||
| # requires the log transformed values in input of intercept | ||
| self$lines <- lapply(lines, log10) |
There was a problem hiding this comment.
If possible could you document the scaling in the @param lines and maybe in the pk-ratio vignette
msevestre
deleted the
add-residualVsObserved-ddi-ratio-guest-function-output
branch
At present the DDI plot Guest bounds are for the predictedVsObserved case.
This PR allows for the residualVsObserved case.