introducing tests for factorial designs

NAMESPACE

@@ -169,6 +169,7 @@ export(summarise_QValues)
 export(summarize_hierarchy)
 export(summarize_stats)
 export(summarize_stats_all)
+export(summarize_stats_factors)
 export(summarize_stats_quantiles)
 export(summary_ROPECA_median_p.scaled)
 export(table_facade)

R/ContrastsSimpleImpute.R

@@ -24,16 +24,38 @@
 #' lProt$rename_response("transformedIntensity")
 #'
 #' Contr <- c("dil.b_vs_a" = "group_A - group_Ctrl")
-#' #ContrastsMissing$debug("get_contrasts")
 #' csi <- ContrastsMissing$new(lProt, contrasts = Contr)
 #' csi$get_contrast_sides()
 #'
 #' res <- csi$get_contrasts()
+#'
 #' stopifnot(nrow(res) ==  (protIntensity$protein_Id |> unique() |> length()))
 #' res$contrast |> table()
 #' stopifnot((res$p.value |> is.na() |> sum()) == 0)
 #' plot(res$diff, -log10(res$p.value), pch = ".")
 #' csi$column_description()
+#' x<- csi$get_Plotter()
+#' p <- x$volcano()
+#' pdf(file = NULL)
+#' print(p)
+#' dev.off()
+#'
+#' dd <- prolfqua::sim_lfq_data_protein_2Factor_config(Nprot = 100,weight_missing = 0.1)
+#'
+#' Contrasts <- c("c1" = "TreatmentA - TreatmentB",
+#'                "C2" = "BackgroundX- BackgroundZ",
+#'                "c3" = "`TreatmentA:BackgroundX` - `TreatmentA:BackgroundZ`",
+#'                "c4" = "`TreatmentB:BackgroundX` - `TreatmentB:BackgroundZ`"
+#'                )
+#' lProt <- LFQData$new(dd$data, dd$config)
+#' lProt$rename_response("transformedIntensity")
+#'
+#' csi <- ContrastsMissing$new(lProt, contrasts = Contrasts)
+#' res <- csi$get_contrasts()
+#' pl <- csi$get_Plotter()
+#' pdf(file = NULL)
+#' pl$volcano()
+#' dev.off()
 ContrastsMissing <- R6::R6Class(
   "ContrastsMissing",
   inherit = ContrastsInterface,

R/simulate_LFQ_data.R

@@ -217,6 +217,7 @@ sim_lfq_data_protein_config <- function(Nprot = 10, with_missing = TRUE, weight_
 #' @param seed seed for reproducibility, if NULL no seed is set.
 #' @export
 #' @examples
+#' undebug(sim_lfq_data_protein_2Factor_config)
 #' x <- sim_lfq_data_protein_2Factor_config()
 #' stopifnot("data.frame" %in% class(x$data))
 #' stopifnot("AnalysisConfiguration" %in% class(x$config))
@@ -228,11 +229,11 @@ sim_lfq_data_protein_2Factor_config <- function(Nprot = 10,
   if (!is.null(seed)) {
     set.seed(seed)
   }
-  res <- sim_lfq_data(Nprot = 10, PEPTIDE = FALSE,
+  res <- sim_lfq_data(Nprot = Nprot, PEPTIDE = FALSE,
                       fc = list(A = c(D = -2,  U = 2, N = 0), B = c(D = 1, U = -4), C = c(D = -1, U = -4)),
                       prop = list(A = c(D = 10, U = 10), B = c(D = 5, U = 20), C = c(D = 15, U = 25)))
   res <- res |> mutate(Treatment = case_when(group %in% c("Ctrl", "A") ~ "A", TRUE ~ "B"))
-  res <- res |> mutate(Background = case_when(group %in% c("Ctrl", "C") ~ "Z", TRUE ~ "X"))
+  data <- res |> mutate(Background = case_when(group %in% c("Ctrl", "C") ~ "Z", TRUE ~ "X"))
   if (with_missing) {
     data <- data[!which_missing(data$abundance,weight_missing = weight_missing),]
   }
@@ -244,6 +245,7 @@ sim_lfq_data_protein_2Factor_config <- function(Nprot = 10,
   atable$nr_children = "nr_peptides"
   atable$factors["Treatment"] = "Treatment"
   atable$factors["Background"] = "Background"
+  atable$factorDepth <- 2
   atable$hierarchy[["protein_Id"]] = c("proteinID", "idtype2")
   atable$set_response("abundance")
 

R/tidyMS_missigness_V2.R

@@ -12,8 +12,20 @@
 #' mh$impute_weighted_lod()
 #' mh$impute_lod()
 #' mh$get_poolvar()
-#' mh$get_contrast_estimates(Contrasts)
+#' bb <- mh$get_contrast_estimates(Contrasts)
 #' mh$get_contrasts(Contrasts)
+#'
+#' dd <- prolfqua::sim_lfq_data_protein_2Factor_config(Nprot = 100,weight_missing = 0.1)
+#'
+#' Contrasts <- c("c1" = "TreatmentA - TreatmentB",
+#'                "C2" = "BackgroundX- BackgroundZ",
+#'                "c3" = "`TreatmentA:BackgroundX` - `TreatmentA:BackgroundZ`",
+#'                "c4" = "`TreatmentB:BackgroundX` - `TreatmentB:BackgroundZ`"
+#'                )
+#' mh <- prolfqua::MissingHelpers$new(dd$data, dd$config, prob = 0.8,weighted = TRUE)
+#' mh$get_stats()$interaction |> table()
+#' mh$get_contrast_estimates(Contrasts)
+#'
 MissingHelpers <- R6::R6Class(
   "MissingHelpers",
 
@@ -44,8 +56,7 @@ MissingHelpers <- R6::R6Class(
     },
     get_stats = function(){
       if (is.null(self$stats)) {
-        self$stats = prolfqua::summarize_stats(self$data, self$config)
-        self$stats = prolfqua::make_interaction_column(self$stats, columns = self$config$table$factor_keys_depth(), sep = ":")
+        self$stats = prolfqua::summarize_stats_factors(self$data, self$config)
       }
       return(self$stats)
     },
@@ -90,6 +101,9 @@ MissingHelpers <- R6::R6Class(
       pooled <- pooled |> mutate(df = ifelse(df == 0, 1, df))
       return(pooled)
     },
+    #' @description
+    #' get contrast estimates
+    #' @param Contrasts named array with contrasts
     get_contrast_estimates = function(
       Contrasts
     ){

R/tidyMS_stats.R

@@ -162,11 +162,28 @@ poolvar <- function(res1, config,  method = c("V1","V2")){
 #'
 #' res1 <- summarize_stats(data, config)
 #'
-summarize_stats <- function(pdata, config){
+#' res2 <- prolfqua::sim_lfq_data_protein_2Factor_config()
+#' res2$config$table$factorDepth <- 2
+#' stats <- summarize_stats(res2$data, res2$config)
+#' stats <- prolfqua::make_interaction_column(stats, columns = res2$config$table$factor_keys_depth(), sep = ":")
+#' stopifnot(nrow(stats) == 40)
+#'
+#' stats <- summarize_stats(res2$data, res2$config, factor_key = res2$config$table$factor_keys()[1])
+#' stats <- prolfqua::make_interaction_column(stats, columns = res2$config$table$factor_keys()[1], sep = ":")
+#' stopifnot(nrow(stats) == 20)
+#' stats <- summarize_stats(res2$data, res2$config, factor_key = res2$config$table$factor_keys()[2])
+#' stats <- prolfqua::make_interaction_column(stats, columns = res2$config$table$factor_keys()[2], sep = ":")
+#' stopifnot(nrow(stats) == 20)
+#'
+#' stats <- summarize_stats(res2$data, res2$config, factor_key = NULL)
+#' stopifnot(nrow(stats) == 10)
+#'
+summarize_stats <- function(pdata, config, factor_key = config$table$factor_keys_depth()){
+  print(factor_key)
   pdata <- complete_cases(pdata, config)
   intsym <- sym(config$table$get_response())
   hierarchyFactor <- pdata |>
-    dplyr::group_by(!!!syms( c(config$table$hierarchy_keys(), config$table$factor_keys_depth()) )) |>
+    dplyr::group_by(!!!syms( c(config$table$hierarchy_keys(), factor_key) )) |>
     dplyr::summarize(nrReplicates = dplyr::n(),
                      nrMeasured = sum(!is.na(!!intsym)),
                      nrNAs = sum(is.na(!!intsym)),
@@ -177,13 +194,46 @@ summarize_stats <- function(pdata, config){
                      .groups = "drop_last") |>  dplyr::ungroup()
 
   hierarchyFactor <- hierarchyFactor |>
-    dplyr::mutate(dplyr::across(config$table$factor_keys_depth(), as.character))
+    dplyr::mutate(dplyr::across(all_of(factor_key), as.character))
   if (config$table$is_response_transformed == FALSE) {
-    hierarchyFactor |> dplyr::mutate(CV = sd/meanAbundance * 100) -> hierarchyFactor
+    hierarchyFactor <- hierarchyFactor |> dplyr::mutate(CV = sd/meanAbundance * 100)
+  }
+  if (is.null(factor_key)) {
+    hierarchyFactor <- dplyr::mutate(hierarchyFactor, !!config$table$factor_keys()[1] := "All")
   }
   return(ungroup(hierarchyFactor))
 }
 
+
+#' compute var sd etc for all factor levels
+#'
+#' @export
+#' @examples
+#' # example code
+#' res2 <- prolfqua::sim_lfq_data_protein_2Factor_config()
+#' xx <- summarize_stats_factors(res2$data, res2$config)
+#' stopifnot(nrow(xx) == 80)
+#'
+summarize_stats_factors <- function(pdata, config){
+  fac_res <- list()
+  stats <- summarize_stats(
+    pdata,
+    config)
+  fac_res[["interaction"]] <- prolfqua::make_interaction_column(stats, columns = config$table$factor_keys_depth(),sep = ":")
+
+  if (config$table$factorDepth > 1 ) { # if 1 only then done
+    for (factor in config$table$factor_keys_depth()) {
+      stats <- summarize_stats(
+        pdata,
+        config,factor_key = factor)
+      fac_res[[factor]]  <- prolfqua::make_interaction_column(stats, columns = factor, sep = ":")
+    }
+  }
+  intfact <- dplyr::bind_rows(fac_res)
+  return(intfact)
+}
+
+
 #' Compute mean, sd, and CV for e.g. Peptides, or proteins, for all samples.
 #'
 #' @param pdata data.frame
@@ -201,28 +251,13 @@ summarize_stats <- function(pdata, config){
 #' res1 <- summarize_stats_all(bb$data, bb$config)
 #'
 #' stopifnot((res1 |> dplyr::filter(group_ == "All") |> nrow()) == (res1 |> nrow()))
-#'
-summarize_stats_all <- function(pdata, config){
-  pdata <- complete_cases(pdata, config)
-  intsym <- sym(config$table$get_response())
-  hierarchy <- pdata |>
-    dplyr::group_by(!!!syms( config$table$hierarchy_keys() )) |>
-    dplyr::summarize(nrReplicates = dplyr::n(),
-                     nrMeasured = sum(!is.na(!!intsym)),
-                     sd = sd(!!intsym,na.rm = TRUE),
-                     var = sd(!!intsym,na.rm = TRUE),
-                     meanAbundance = mean(!!intsym,na.rm = TRUE),
-                     medianAbundance = median(!!intsym, na.rm = TRUE),
-                     .groups = "drop_last") |> dplyr::ungroup()
-
-  hierarchy <- dplyr::mutate(hierarchy, !!config$table$factor_keys()[1] := "All")
-  hierarchyFactor <- hierarchy
-  if (config$table$is_response_transformed == FALSE) {
-    hierarchyFactor |> dplyr::mutate(CV = sd/meanAbundance * 100) -> hierarchyFactor
-  }
-  return(ungroup(hierarchyFactor))
+#' res2 <- prolfqua::sim_lfq_data_protein_2Factor_config()
+#' resSt <- summarize_stats_all(res2$data, res2$config)
+summarize_stats_all <- function(pdata, config) {
+  summarize_stats(pdata, config, factor_key = NULL)
 }
 
+
 #' summarize stats output (compute quantiles)
 #' @param stats_res result of running `summarize_stats`
 #' @param config AnalysisConfiguration
@@ -247,7 +282,6 @@ summarize_stats_all <- function(pdata, config){
 #' stats_res <- summarize_stats(data, config)
 #' sq <- summarize_stats_quantiles(stats_res, config)
 #' sq <- summarize_stats_quantiles(stats_res, config, stats = "sd")
-#'
 #' stats_res <- summarize_stats(data, config)
 #' xx <- summarize_stats_quantiles(stats_res, config, probs = seq(0,1,by = 0.1))
 #' ggplot2::ggplot(xx$long, aes(x = probs, y = quantiles, color = group_)) + geom_line() + geom_point()