add randomforest

R/impute.R

@@ -100,10 +100,48 @@ impute <- function(data,
                    comparison = comparison,
                    missingness = missingness,
                    noise = NULL,
-                   method = "ludovic",
+                   method = c("ludovic", "randomforest"),
                    skip_log2_transform_error = FALSE,
-                   retain_columns = NULL) {
+                   retain_columns = NULL,
+                   ...) {
   noise_missing <- missing(noise) # check if argument noise was provided or not
+
+  if (method == "randomforest"){
+    data_wide <- data %>%
+      tidyr::pivot_wider(
+        id_cols=sample,
+        names_from =grouping,
+        values_from = intensity_log2) %>%
+      dplyr::select(sample)
+
+    data_wide <- data.frame(lapply(data_wide, function(x) as.numeric(as.character(x))))
+    data_imputed_rf <- missForest(data_wide, ...)
+
+    data_imputed <-data_imputed_rf$ximp
+    data_imputed[sample] <- data %>%
+      tidyr::pivot_wider(
+        id_cols = sample,
+        names_from = grouping,
+        values_from = intensity_log2) %>%
+      dplyr::pull(sample)
+
+    data_imputed <- data_imputed %>%
+      as.data.frame() %>%
+      tidyr::pivot_longer(
+        cols = -sample,
+        names_to = grouping,
+        values_to = intensity_log2
+      )
+
+    result <- data_imputed %>%
+      left_join(
+        synthetic_data %>%
+          dplyr::select(sample, retain_columns) %>%
+          distinct(),
+        by = sample
+      )
+    return(result)
+  }
   result <- data %>%
     dplyr::distinct(
       {{ sample }},