Bug: AUC start 0 (which also needs Feature: C0 imputation) (#138)

* bug: force doses times to be AFRLT.dose (at ARRLT = 0) * data: correct PKROUTE -> ROUTE * enhancement: create function to create impute_c0 column * enhancement: include additional C0 imputation strategies * feature: intervals include imputate strategy with duplicated rows apart for non-c0 parameters * bug: PKNCAconc formula adapted for PKNCA to consider predoses for impute_predose * bug: solved classification of doses in create_c0_impute * bug: perfectionated formulas to consider different analytes appropiately * style: solve most lintr issues * refactor: delete report files that should not have been called * refactor: nolint in customized PKNCA_impute_method functions * wordlist: add PKNCA_impute_method functions * documentation: functions in R/PKNCA_impute_method_additions.R * feature: c0 input & when no checked use as interval start 1st observation * bug: solved PKNCA warnings "No concentration data" * bug: solved issue of duplicated tmax in results * refactor: NCA results with all dose and conc columns first * bug: fixed impute_method_start_c1 unneded argument issue * refactor: delete unused functions in reshape * refactor: input selectors for analyte and dosno put together in UI and renamed (select_X) * refactor: change strategy for input of AUC intervals * bug: fixed new intervals strategy and align with PKNCA running * revert feat: back to old UI function for AUC intervals * fix: interval default start since dose, aucint for manual intervals, adjust reshape for it * fix: when C0 is not imputed, start is at C1 * lintr: code cleaning * fix: issue in slope_selector with new name for input$cyclenca * refactor: make own function to create intervals with dose data * fix: adapt lambda_slope_plot to use type_interval column instead of end==Inf * fix: adapt create_c0_impute_column to not use artificial time_dose column * fix bug: correct call for mydata$dose instead of mydose * fix: solve issues when C0 is not imputed to start at C1 * lintr: lint_dir cleaning * fix oldrel: deleted line used for testing * fix: update roxygen2 documentation for format_data.R functions * fix: spellcheck added words to WORDLIST * fix: roxygen2 for missing documetnation PKNCA_impute_method_additions.R * fix: solved issue app crashes with auc rule sets * fix: created pknca_trigger to run slope_selector when apply/save changes * refactor: data selection tab with arranged dataset by subject * aesthetics: add hovering message and new label in checkbox C0 imputations * fix: lintr, roxygen doc and clast added to default params in NCA * fix: reporting all parameters instead of just auc related for intervals * documentation: roxygen update * fix: delete partial_auc_input.R from R/ folder and leave in inst/shiny/functions * fix: add tests and prevent input errors in format_data.R functions * fix: issue with select_analyte after merge * test: add tests for PKNCA_impute_method_additions.R and prevent NA in start_log * refactor: create_c0_impute testing, column preventions and renaming (create_start_impute) * fix: lintr code cleaning and roxygen documentation * fix: update description and wordlist * fix: impossible test deleted and improved results column order display * style: add details to function description and rename PKNCA_impute_method_start_log(slope) * style: time == start & link on PKNCA custom function guidelines * test: modify tests from expect_s3_class to expect_no_error for nested functions * style: delete unnecessary variable assignment in function * documentation: update roxygen2 * fix: filtering later and using dose.time + arrlt for non-t0 solves partially problem * data: corect ADOSEDUR (ADOSEDUR<0 turned into |ADOSEDUR|) * data: adapt USUBJID=11102, DOSNO=9 to a reasonable dose time (AFRLT-ARRLT = 1680.5) adnca2 %>% mutate(AFRLT = ifelse(DOSNO==9 & USUBJID==11102, AFRLT + 168, AFRLT), NFRLT = ifelse(DOSNO==9 & USUBJID==11102, NFRLT + 168, NFRLT) ) * fix: solve issue with end<start due to incorrect create_dose rows output * refactor: lintr code cleaning * style: apply suggestions from m-kolomanski Co-authored-by: Mateusz Kołomański <[email protected]> * style: drug_column call instead of hardcoded "DRUG" in create_start_impute.R * style: lintr code cleaning * style: improve conditional syntax in create_start_impute.R * style issue: clean computation of pk.calc.c0 logslope * style: suggestions from code review m-kolomanski Co-authored-by: Mateusz Kołomański <[email protected]> * nitpick: solve nitpicks pointed out by m-kolomanski * refactor: rename reshape_PKNCA_results to pivot_wider_pknca_results * refactor: change the names of format_data.R functions and update documentation * fix: format_pkncadose_data includes either DOSNO or TIME (dose time) to slice data. route_column (unneeded argument) was removed * refactor: lintr cleaning and roxygen2 documentation update * spelling: update wordlist * suggestions: nitpicks, returns m-kolomanski Co-authored-by: Mateusz Kołomański <[email protected]> * refactor: clarify with logical columns create_start_impute.R assignation * issue: eliminate all redundant dplyr/tidyr namespacing * issue: change base function to dplyr (base::merge > dplyr::inner_join) * fix: use describe/it in tests PKNCA_impute_method_additions & create_start_impute * nitpick, suggestion: nca.R add backlash and de-namespace mutate Co-authored-by: Mateusz Kołomański <[email protected]> * fix: filtering of dosno/analyte in displayed in `Data Selection` tab table * fix: crash on slope_selector due to filtering results with NA lambda.z * style: lintr clean code * style: lintr code cleaning * fix: remove redundant inst/WORDLIST.txt --------- Co-authored-by: Mateusz Kołomański <[email protected]>
pharmaverse · Jan 17, 2025 · 64dc3dc · 64dc3dc
1 parent 430c969
commit 64dc3dc
Show file tree

Hide file tree

Showing 25 changed files with 1,816 additions and 1,049 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,5 +1,5 @@
 Package: aNCA
-Title: (Pre-)clinical NCA in a dynamic shiny app
+Title: (Pre-)Clinical NCA in a Dynamic Shiny App
 Version: 0.0.0.9000
 Authors@R: c(
     person("Ercan", "Suekuer", , "[email protected]", role = c("aut", "cre"),
@@ -62,6 +62,7 @@ Imports:
     tidyr,
     tools,
     utils,
+    rlang,
     yaml,
     zip
 Encoding: UTF-8

diff --git a/NAMESPACE b/NAMESPACE
@@ -1,15 +1,18 @@
 # Generated by roxygen2: do not edit by hand
 
+export(PKNCA_impute_method_start_c1)
+export(PKNCA_impute_method_start_logslope)
 export(anonymize_pk_data)
 export(apply_filters)
 export(apply_labels)
 export(as_factor_preserve_label)
 export(calculate_summary_stats)
-export(create_conc)
-export(create_dose)
+export(create_start_impute)
 export(filter_breaks)
 export(flexible_violinboxplot)
-export(format_data)
+export(format_pkncaconc_data)
+export(format_pkncadata_intervals)
+export(format_pkncadose_data)
 export(g_pkconc_ind_lin)
 export(g_pkconc_ind_log)
 export(general_lineplot)
@@ -18,11 +21,11 @@ export(geometric_mean)
 export(get_label)
 export(has_label)
 export(lambda_slope_plot)
+export(pivot_wider_pknca_results)
 export(parse_annotation)
 export(parse_tlg_definitions)
 export(pkcg01)
 export(pptestcd_dict)
-export(reshape_pknca_results)
 export(run_app)
 export(set_empty_label)
 import(bslib)
@@ -43,6 +46,7 @@ importFrom(PKNCA,PKNCA.options)
 importFrom(PKNCA,PKNCAconc)
 importFrom(PKNCA,PKNCAdata)
 importFrom(PKNCA,PKNCAdose)
+importFrom(PKNCA,pk.calc.c0)
 importFrom(PKNCA,pk.nca)
 importFrom(PKNCA,pknca_units_table)
 importFrom(checkmate,assert_numeric)
@@ -99,6 +103,7 @@ importFrom(reactable,renderReactable)
 importFrom(reactable.extras,dropdown_extra)
 importFrom(reactable.extras,text_extra)
 importFrom(rio,export_list)
+importFrom(rlang,sym)
 importFrom(rmarkdown,render)
 importFrom(scales,breaks_log)
 importFrom(scales,label_log)

diff --git a/R/PKNCA_impute_method_additions.R b/R/PKNCA_impute_method_additions.R
@@ -0,0 +1,75 @@
+#' This function imputes the start concentration using the log slope method.
+#'
+#' @param conc Numeric vector of concentrations.
+#' @param time Numeric vector of times corresponding to the concentrations.
+#' @param start Numeric value indicating the start/dose time.
+#' @param end Numeric value indicating the end time.
+#' @param ... Additional arguments (currently not used).
+#' @param options List of options (currently not used).
+#'
+#' @returns A data frame with imputed start concentration.
+#' @details
+#' This function adheres to the structure required by the `PKNCA` package to work with its
+#' functionalities. For more information, see the
+#' [PKNCA Data Imputation Vignette](https://cran.r-project.org/web/packages/PKNCA/vignettes).
+#' @export
+#'
+#' @examples
+#' conc <- c(5, 4, 3, 2, 1)
+#' time <- c(1, 2, 3, 4, 5)
+#' start <- 0
+#' end <- 4
+#' PKNCA_impute_method_start_logslope(conc, time, start, end)
+
+PKNCA_impute_method_start_logslope <- function(conc, time, start, end, ..., options = list()) { # nolint
+
+  d_conc_time <- data.frame(conc = conc, time = time)
+  if (!any(time == start)) {
+    all_concs <- conc[time >= start  &  time <= end]
+    all_times <- time[time >= start  &  time <= end]
+    if (!all(is.na(all_concs))) {
+      c0 <- PKNCA::pk.calc.c0(all_concs, all_times, method = "logslope")
+      if (!is.na(c0)) {
+        d_conc_time <- rbind(d_conc_time, data.frame(time = start, conc = c0))
+        d_conc_time <- d_conc_time[order(d_conc_time$time), ]
+      }
+    }
+  }
+  d_conc_time
+}
+
+#' This function imputes the start concentration using the first concentration after dose
+#'
+#' @param conc Numeric vector of concentrations.
+#' @param time Numeric vector of times corresponding to the concentrations.
+#' @param start Numeric value indicating the start/dose time.
+#' @param end Numeric value indicating the end time.
+#' @param ... Additional arguments (currently not used).
+#' @param options List of options (currently not used).
+#'
+#' @returns A data frame with imputed start concentration.
+#' @details
+#' This function adheres to the structure required by the `PKNCA` package to work with its
+#' functionalities.For more information, see the
+#' [PKNCA Data Imputation Vignette](https://cran.r-project.org/web/packages/PKNCA/vignettes).
+#' @export
+#'
+#' @examples
+#' conc <- c(1, 2, 3, 4, 5)
+#' time <- c(1, 2, 3, 4, 5)
+#' start <- 0
+#' end <- 4
+#' PKNCA_impute_method_start_c1(conc, time, start, end)
+PKNCA_impute_method_start_c1 <- function(conc, time, start, end, ..., options = list()) { # nolint
+  d_conc_time <- data.frame(conc = conc, time = time)
+  if (!any(time == start)) {
+    all_concs <- conc[time >= start & time <= end]
+    all_times <- time[time >= start & time <= end]
+    if (!all(is.na(all_concs))) {
+      c1 <- all_concs[which.min(all_times)]
+      d_conc_time <- rbind(d_conc_time, data.frame(time = start, conc = c1))
+      d_conc_time <- d_conc_time[order(d_conc_time$time), ]
+    }
+  }
+  d_conc_time
+}
diff --git a/R/create_start_impute.R b/R/create_start_impute.R
@@ -0,0 +1,101 @@
+#' Create C0 Impute Column
+#'
+#' Defines an impute column in the intervals of the PKNCAdata object based on data
+#'
+#' @param mydata A PKNCAdata object containing concentration and dose data.
+#' @returns A PKNCAdata object with updated intervals table including start imputation strategies.
+#' @import dplyr
+#' @importFrom rlang sym
+#' @importFrom PKNCA pk.calc.c0
+#' @export
+#'
+#' @examples
+#' \dontrun{
+#' mydata <- create_start_impute(mydata)
+#' }
+create_start_impute <- function(mydata) {
+
+  # Define columns
+  conc_column <- mydata$conc$columns$conc
+  time_column <- mydata$conc$columns$time
+  analyte_column <- mydata$conc$columns$groups$group_analyte
+  route_column <- mydata$dose$columns$route
+  duration_column <- mydata$dose$columns$duration
+  drug_column <- "DRUG"  # TODO: At some point should be mapped manually in the App
+  conc_group_columns <- unname(unlist(mydata$conc$columns$groups))
+  dose_group_columns <- unname(unlist(mydata$dose$columns$groups))
+  group_columns <- unique(c(conc_group_columns, dose_group_columns))
+
+  # Define dose number (DOSNO) if not present in dose data
+  if (!"DOSNO" %in% mydata$dose$data) {
+    mydata$dose$data <- mydata$dose$data %>%
+      group_by(across(all_of(dose_group_columns))) %>%
+      mutate(DOSNO = row_number()) %>%
+      ungroup()
+  }
+
+  mydata_with_int <- merge(
+    x = mydata$conc$data %>%
+      select(any_of(c(conc_group_columns, conc_column, time_column))),
+    y = mydata$dose$data %>%
+      select(any_of(c(dose_group_columns, route_column,
+                      duration_column, "DOSNO", "DRUG")))
+  ) %>%
+    merge(mydata$intervals)
+
+  # Define dosing drug as analyte if not present
+  if (!drug_column %in% colnames(mydata_with_int)) {
+    if (analyte_column %in% colnames(mydata_with_int)) {
+      drug_column <- analyte_column
+    } else {
+      mydata_with_int <- mutate(mydata_with_int,
+                                ANALYTE = "A")
+      analyte_column <- "ANALYTE"
+      drug_column <- analyte_column
+    }
+  }
+
+  # Process imputation strategy based on each interval
+  mydata$intervals <- mydata_with_int %>%
+    group_by(across(all_of(c(group_columns, "DOSNO", "start", "end")))) %>%
+    arrange(across(all_of(c(group_columns, time_column)))) %>%
+    mutate(
+      is.first.dose = DOSNO == 1,
+      is.ivbolus = tolower(!!sym(route_column)) == "intravascular" & !!sym(duration_column) == 0,
+      is.analyte.drug = !!sym(analyte_column) == !!sym(drug_column),
+      is.possible.c0.logslope = !is.na(pk.calc.c0(conc = !!sym(conc_column),
+                                                  time = !!sym(time_column),
+                                                  time.dose = start[1],
+                                                  method = "logslope"))
+    ) %>%
+    arrange(
+      (!!sym(time_column) - start) < 0,
+      (!!sym(time_column) - start)
+    ) %>%
+    slice(1) %>%
+    ungroup() %>%
+    rowwise() %>%
+    mutate(
+      impute = case_when(
+        # Start concentration already present: No imputation (NA)
+        !!sym(time_column) == start & !is.na(!!sym(conc_column)) ~ NA,
+
+        # 1st dose with not IV bolus or analyte =! drug : Start concentration is 0
+        is.first.dose & (!is.ivbolus | !is.analyte.drug) ~ "start_conc0",
+
+        # Posterior doses not IV bolus or analyte =! drug : Start concentration shifts to predose
+        !is.first.dose & (!is.ivbolus | !is.analyte.drug) ~ "start_predose",
+
+        # IV bolus with analyte = drug : Start concentration is log-backextrapolated (if possible)
+        is.ivbolus & is.analyte.drug & is.possible.c0.logslope ~ "start_logslope",
+
+        # IV bolus with analyte = drug and not possible logslope: Start concentration is 1st conc
+        is.ivbolus & is.analyte.drug ~ "start_c1"
+      )
+    ) %>%
+    # Select only the columns of interest
+    select(any_of(c(names(mydata$intervals), "impute")))
+
+  mydata
+
+}