Release v 2.0.1

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: familiar
 Title: End-to-End Automated Machine Learning and Model Evaluation
-Version: 2.0.0
+Version: 2.0.1
 Authors@R: c(
     person("Alex", "Zwanenburg",
            email = "alexander.zwanenburg@nct-dresden.de",
@@ -26,7 +26,6 @@ Depends: R (>= 4.0.0)
 License: EUPL
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.3.3
 Roxygen: list(markdown = TRUE)
 VignetteBuilder: knitr
 Imports:
@@ -227,3 +226,4 @@ Collate:
     'aaa.R'
 Config/testthat/parallel: true
 Config/testthat/edition: 3
+Config/roxygen2/version: 8.0.0

NAMESPACE

@@ -70,6 +70,7 @@ exportMethods(plot_feature_similarity)
 exportMethods(plot_ice)
 exportMethods(plot_kaplan_meier)
 exportMethods(plot_model_performance)
+exportMethods(plot_pd)
 exportMethods(plot_permutation_variable_importance)
 exportMethods(plot_sample_clustering)
 exportMethods(plot_shap_dependence)

NEWS.md

@@ -1,3 +1,35 @@
+# Version 2.0.1 (Brilliant Bat)
+
+## Minor changes
+
+- `plot_feature_similarity` and `plot_sample_clustering` now have the `features` 
+  argument, which selects the features that appear in the heatmap.
+
+- `plot_feature_similarity` and `plot_sample_clustering` now have the 
+  `remove_feature_labels` argument, which suppresses rendering of feature 
+  names as tick labels for the heatmap.
+
+- `plot_sample_clustering` now has the `remove_sample_labels` argument, which
+  suppresses rendering of sample names as tick labels for the heatmap.
+
+- `plot_kaplan_meier` now plots strata in one facet if the risk group is equal
+  to the data set.
+
+- `plot_kaplan_meier` now correctly shows the results of the log-rank test for
+  strata in each facet.
+
+## Fixes
+
+- Prevents an error when evaluating models in bootstrap or cross-validation
+  experiments without variable importances and a number features in the model
+  that exceeds `n_important_features`.
+
+- `plot_pd` is now correctly exported and can be called externally.
+
+- `plot_kaplan_meier` now works correctly if the column indicated by `risk_group_column`
+  contains `NA` values.
+
+
 # Version 2.0.0 (Astonishing Anteater)
 
 ## Breaking changes

R/DataObject.R

@@ -2338,7 +2338,10 @@ setMethod(
     # Check if features are present as column name
     if (length(features) > 0L) {
       if (!all(features %in% colnames(data@data))) {
-        logger_stop("Not all features were found in the data set.")
+        missing_features <- setdiff(features, colnames(data@data))
+        logger_stop(
+          paste0("Not all features were found in the data set: ", paste_s(missing_features))
+        )
       }
     }
 

R/DataParameterChecks.R

@@ -73,6 +73,11 @@
     ) {
       settings$data$sample_col <- get_id_columns(single_column = "sample")
     }
+
+  } else if (is.null(settings$data$sample_col)) {
+    if (get_id_columns(single_column = "sample") %in% colnames(data)) {
+      settings$data$sample_col <- get_id_columns(single_column = "sample")
+    }
   }
 
   if (!is.null(settings$data$batch_col) && check_stringency != "strict") {
@@ -83,6 +88,11 @@
     ) {
       settings$data$batch_col <- get_id_columns(single_column = "batch")
     }
+
+  } else if (is.null(settings$data$batch_col)) {
+    if (get_id_columns(single_column = "batch") %in% colnames(data)) {
+      settings$data$batch_col <- get_id_columns(single_column = "batch")
+    }
   }
 
   if (!is.null(settings$data$series_col) && check_stringency != "strict") {
@@ -93,6 +103,11 @@
     ) {
       settings$data$series_col <- get_id_columns(single_column = "series")
     }
+
+  } else if (is.null(settings$data$series_col)) {
+    if (get_id_columns(single_column = "series") %in% colnames(data)) {
+      settings$data$series_col <- get_id_columns(single_column = "series")
+    }
   }
 
   ## Sample column -------------------------------------------------------------

R/Familiar.R

@@ -609,8 +609,7 @@ summon_familiar <- function(
 #' @inheritParams .parse_variable_importance_settings
 #' @inheritDotParams .parse_experiment_settings -config
 #' @inheritDotParams .parse_setup_settings -config
-#' @inheritDotParams .parse_preprocessing_settings -config -data -parallel
-#'   -outcome_type
+#' @inheritDotParams .parse_preprocessing_settings -config -data -parallel -outcome_type
 #'
 #' @details This is a thin wrapper around `summon_familiar`, and functions like
 #'   it, but automatically skips computation of variable importance, learning
@@ -729,8 +728,7 @@ precompute_data_assignment <- function(
 #' @inheritParams .parse_variable_importance_settings
 #' @inheritDotParams .parse_experiment_settings -config
 #' @inheritDotParams .parse_setup_settings -config
-#' @inheritDotParams .parse_preprocessing_settings -config -data -parallel
-#'   -outcome_type
+#' @inheritDotParams .parse_preprocessing_settings -config -data -parallel -outcome_type
 #'
 #' @details This is a thin wrapper around `summon_familiar`, and functions like
 #'   it, but automatically skips computation of variable importance, learning

R/FamiliarDataComputation.R

@@ -202,6 +202,11 @@ setMethod(
 #'  vignette on performance metrics for the available metrics. If not provided
 #'  explicitly, this parameter is read from settings used at creation of the
 #'  underlying `familiarModel` objects.
+#'
+#'@param features Features that should be considered for extracting information
+#'  from. Typically called in external workflows, e.g. for plotting. Internally,
+#'  i.e. from summon_familiar, this variable is not used.
+#'
 #'@param feature_cluster_method The method used to perform clustering. These are
 #'  the same methods as for the `cluster_method` configuration parameter:
 #'  `none`, `hclust`, `agnes`, `diana` and `pam`.
@@ -319,6 +324,7 @@ setMethod(
     stratification_method = waiver(),
     evaluation_times = waiver(),
     metric = waiver(),
+    features = waiver(),
     feature_cluster_method = waiver(),
     feature_cluster_cut_method = waiver(),
     feature_linkage_method = waiver(),
@@ -346,7 +352,7 @@ setMethod(
             "feature_expressions", "permutation_vimp") %in% data_element)) {
     # Not for the fs_vimp data elements. This is because the subset of
     # features in the ensemble is generally smaller than that assessed within
-    # feature selection..
+    # feature selection.
 
     # Compute a table containing the pairwise distance between features.
     feature_similarity <- extract_feature_similarity(
@@ -358,6 +364,7 @@ setMethod(
       confidence_level = confidence_level,
       bootstrap_ci_method = bootstrap_ci_method,
       is_pre_processed = is_pre_processed,
+      features = features,
       feature_cluster_method = feature_cluster_method,
       feature_linkage_method = feature_linkage_method,
       feature_cluster_cut_method = feature_cluster_cut_method,
@@ -378,6 +385,7 @@ setMethod(
       data = data,
       cl = cl,
       is_pre_processed = is_pre_processed,
+      features = features,
       sample_limit = sample_limit,
       sample_similarity_metric = sample_similarity_metric,
       sample_cluster_method = sample_cluster_method,
@@ -639,6 +647,7 @@ setMethod(
       data = data,
       cl = cl,
       ensemble_method = ensemble_method,
+      features = features,
       evaluation_times = evaluation_times,
       sample_limit = sample_limit,
       n_important_features = n_important_features,
@@ -663,6 +672,7 @@ setMethod(
       data = data,
       cl = cl,
       ensemble_method = ensemble_method,
+      features = features,
       evaluation_times = evaluation_times,
       sample_limit = sample_limit,
       detail_level = detail_level,

R/FamiliarDataComputationFeatureExpression.R

@@ -279,13 +279,37 @@ setMethod(
 
 
 
+
+.feature_expression_remove_feature <- function(x, features) {
+  if (is_empty(x)) return(x)
+
+  all_features <- names(x@feature_info)
+  available_features <- intersect(all_features, features)
+
+  if (length(available_features) == 0L) return(NULL)
+
+  # Filter features.
+  x@feature_info <- x@feature_info[available_features]
+  x@data <- x@data[, mget(c(setdiff(colnames(x@data), all_features), available_features))]
+  x@value_column <- available_features
+
+  return(x)
+}
+
+
+
+
+
 # export_feature_expressions (generic) -----------------------------------------
 
 #'@title Extract and export feature expressions.
 #'
 #'@description Extract and export feature expressions for the features in a
 #'  familiarCollection.
 #'
+#'@param features Features that should be exported. If `NULL` or `waiver()`, all
+#'  features exported (default).
+#'
 #'@param evaluation_time One or more time points that are used to create the
 #'  outcome columns in expression plots. If not provided explicitly, this
 #'  parameter is read from settings used at creation of the underlying
@@ -314,6 +338,7 @@ setGeneric(
   "export_feature_expressions",
   function(
     object,
+    features = waiver(),
     dir_path = NULL,
     evaluation_time = waiver(),
     export_collection = FALSE,
@@ -333,6 +358,7 @@ setMethod(
   signature(object = "familiarCollection"),
   function(
     object,
+    features = waiver(),
     dir_path = NULL,
     evaluation_time = waiver(),
     export_collection = FALSE,
@@ -348,6 +374,11 @@ setMethod(
     # Check that the data are not empty.
     if (is_empty(x)) return(NULL)
 
+    if (!is.waive(features) && !is.null(features)) {
+      # Filter features.
+      x <- lapply(x, .feature_expression_remove_feature, features = features)
+    }
+
     if (!is.waive(evaluation_time)) {
 
       # Check values.
@@ -391,6 +422,7 @@ setMethod(
   signature(object = "ANY"),
   function(
     object,
+    features = waiver(),
     dir_path = NULL,
     evaluation_time = waiver(),
     export_collection = FALSE,
@@ -403,6 +435,7 @@ setMethod(
       args = c(
         list(
           "object" = object,
+          "features" = features,
           "data_element" = "feature_expressions",
           "evaluation_times" = evaluation_time
         ),
@@ -415,6 +448,7 @@ setMethod(
       args = c(
         list(
           "object" = object,
+          "features" = features,
           "dir_path" = dir_path,
           "evaluation_time" = evaluation_time,
           "export_collection" = export_collection