fix bugs

Author: stemangiola

R/dplyr_methods.R

@@ -62,7 +62,7 @@ bind_rows.SummarizedExperiment <- function(..., .id=NULL, add.cell.ids=NULL) {
 bind_cols_internal <- function(..., .id=NULL, column_belonging=NULL) {
     tts <- tts <- flatten_if(dots_values(...), is_spliced)
 
-<<<<<<< HEAD
+
     # If I have column corresponding bind directly
     # Without tranformation to tibble
     if(!is.null(column_belonging)){
@@ -683,19 +683,28 @@ select.SummarizedExperiment <- function(.data, ...) {
         data.frame(row.names=pull(., !!f_(.data)$symbol)) |>
         select(-!!f_(.data)$symbol) |>
         DataFrame()
-  
+    
+    # If SE does not have rownames, 
+    # I have to take them our of here, otherwise count integration, 
+    # which is a matrix and behaved differently from DataFrame fails
+    if(rownames(.data) |> is.null()) rownames(row_data_DF)  = NULL
+    
     col_data_tibble <- 
         colData(.data) |> 
         as_tibble(rownames = s_(.data)$name) 
 
     col_data_DF <-
         col_data_tibble |>  
-        select(one_of(columns_query), !!s_(.data)$symbol) |>
-        suppressWarnings() %>% 
-        data.frame(row.names=pull(., !!s_(.data)$symbol)) |>
+        select(any_of(columns_query), !!s_(.data)$symbol) |>
+        data.frame(row.names=pull(col_data_tibble, !!s_(.data)$symbol)) |>
         select(-!!s_(.data)$symbol) |>
         DataFrame()
 
+    # If SE does not have rownames, 
+    # I have to take them our of here, otherwise count integration, 
+    # which is a matrix and behaved differently from DataFrame fails
+    if(colnames(.data) |> is.null()) rownames(col_data_DF)  = NULL
+    
     count_data <-
         assays(.data)@listData %>%
             .[names(assays(.data)@listData) %in% columns_query]

R/tidyr_methods.R

@@ -219,6 +219,14 @@ nest.SummarizedExperiment <- function(.data, ..., .names_sep=NULL) {
         .data <- ping_old_special_column_into_metadata(.data)
     }
 
+    # if the data does not have raw, names or clumn names
+    # we have to add them, otherwise the nesting and a nesting will get confused with 
+    # the link between Sample wise, columns, and Sample IDs
+    if(rownames(.data) |> is.null() | colnames(.data) |> is.null() )
+      warning("tidySummarizedExperiment says: the nesting and unnesting operations require row names and column names to avoid side-effects. Therefore, doors will be added as \"1\", \"2\", \"3\".")
+    if(rownames(.data) |> is.null() ) rownames(.data) = .data |> nrow() |> seq_len() |> as.character()
+    if(colnames(.data) |> is.null() ) colnames(.data) = .data |> ncol() |> seq_len() |> as.character()
+    
     my_data__ <- .data 
 
     # Names
@@ -256,12 +264,6 @@ nest.SummarizedExperiment <- function(.data, ..., .names_sep=NULL) {
             " We are working for optimising a generalised solution of nest().")
     }
 
-    # my_data__nested <-
-    #     my_data__ %>% 
-    #     # This is needed otherwise nest goes into loop and fails
-    #     as_tibble() %>%
-    #     tidyr::nest(...)
-     
     # If I nest only for .feature -> THIS WORKS ONLY WITH THE CHECK ABOVE
     if (feature_name %in% colnames(my_test_nest)) {
         return(
@@ -280,7 +282,7 @@ nest.SummarizedExperiment <- function(.data, ..., .names_sep=NULL) {
 
 
     my_data__ %>%
-        select(!!sample_symbol, !!feature_symbol, my_nesting_column) |> 
+        select(!!sample_symbol, !!feature_symbol, all_of(my_nesting_column)) |> 
         as_tibble() %>%
         tidyr::nest(...) |> 
 
@@ -311,15 +313,18 @@ nest.SummarizedExperiment <- function(.data, ..., .names_sep=NULL) {
                     # Here I am filtering because if I have 0 features this leads to failure
                     else my_transcripts= ..1 |> filter(!is.na(!!feature_symbol)) |>  pull(!!feature_symbol)
 
-                    ###
-
+                    # if the summarised experiment does not have feature ID or Sample ID, 
+                    # convert back those to integers
+                    # Note to self: this is convoluted, because if feature ID and sample ID 
+                    # were to stay integers all along, these will not be needed
+                    if(rownames(my_data__) |> is.null()) my_transcripts = as.integer(my_transcripts)
+                    if(colnames(my_data__) |> is.null()) my_samples = as.integer(my_samples)
+                    
                     my_data__[unique(my_transcripts),unique(my_samples)] |>
-                      select(-one_of(
+                      select(-any_of(
                         my_nesting_column |> 
                           setdiff(c(sample_name, feature_name))
-                      )) |> 
-                      suppressWarnings()
-
+                      )) 
 
                 }
             )

R/utilities.R

@@ -749,7 +749,8 @@ get_count_datasets <- function(se) {
   se <- check_se_dimnames(se)
 
   # Join assays
-  map2( 
+  list_assays = 
+    map2( 
     assays(se, withDimnames = FALSE) %>% as.list(),
     names(assays(se)),
     ~ {
@@ -807,21 +808,27 @@ get_count_datasets <- function(se) {
       f_(se)$name %in% colnames(.x) %>% not ~ mutate(.x, !!f_(se)$symbol := as.character(NA)),
       s_(se)$name %in% colnames(.x) %>% not ~ mutate(.x, !!s_(se)$symbol := as.character(NA)),
       ~ .x
-    )) |> 
-    
-    when(
-      length(.)>0 ~ 
-        
-        reduce(., full_join, by = c(f_(se)$name, s_(se)$name)),
-      # reduce(., left_join, by = c(f_(se)$name, s_(se)$name)),
-      # bind_cols(.,  .name_repair = c("minimal")) %>% .[!duplicated(colnames(.))], 
-      ~ expand.grid(
-        rownames(se), colnames(se)
-      ) %>% 
-        setNames(c(f_(se)$name, s_(se)$name)) %>%
-        tibble::as_tibble()
-    ) 
+    )) 
+  
+  # If assays is non empty 
+  if(list_assays |> length() > 0)
+    list_assays |> 
+    reduce(full_join, by = c(f_(se)$name, s_(se)$name))
 
+  # If assays is empty 
+  else {
+    
+    # If I don't have row column names
+    if(se |> rownames() |> is.null()) rn = nrow(se) |> seq_len() |> as.character()
+    else rn = rownames(se)
+    if(se |> colnames() |> is.null()) cn = ncol(se) |> seq_len() |> as.character()
+    else cn = colnames(se)
+    
+    expand.grid(  rn, cn  ) |> 
+             setNames(c(f_(se)$name, s_(se)$name)) |> 
+             tibble::as_tibble()
+  }
+   
 
 }