src/05_DESEQ2_normalisation.R: create correlation figure for t3

src/05_DESEQ2_normalisation.R

@@ -28,6 +28,10 @@ load_n_filter_data <- function(sample_threshold = 5e6, gene_threshold = 10,
             c(1, grep(pattern, colnames(full_data), value = F, perl = T))
         ]
     }
+    coding_gene <- read_tsv(
+            "results/coding_genes/coding_gene.txt"
+        )$gene_id # nolint
+    full_data <- full_data[full_data$gene %in% coding_gene, ]
     if (sample_threshold > 0) {
         # filter by read number
         sample_size <- full_data %>%
@@ -43,17 +47,15 @@ load_n_filter_data <- function(sample_threshold = 5e6, gene_threshold = 10,
             "./results/matrice_count/ercc_matrices.txt"
         ) # nolint
         ercc_data <- ercc_data[, colnames(full_data)]
-        full_data <- rbind(full_data, ercc_data)
-    }
         if (gene_threshold > 0) {
-        full_data$rmean <- full_data %>%
+            ercc_data$rmean <- ercc_data %>%
                 select(-gene) %>%
                 apply(1, mean) # nolint
-        full_data <- full_data %>% filter(rmean >= gene_threshold) # nolint
-        full_data <- full_data %>% select(-rmean) # nolint
+            ercc_data <- ercc_data %>% filter(rmean >= gene_threshold) # nolint
+            ercc_data <- ercc_data %>% select(-rmean) # nolint
         }
-    if (load_ercc) {
-        ercc_gene <- full_data %>%
+        full_data <- rbind(full_data, ercc_data)
+        ercc_gene <- ercc_data %>%
             filter(startsWith(gene, "ERCC-")) %>%
             select(gene) %>%
             unlist() %>%
@@ -111,8 +113,9 @@ normalise_matrix <- function(stable_vector, count_tibble) {
 #' @param output_f Folder where the figures will be created
 #' @param output_file The name of the file that will be created
 #' @param color_col The column to use to color the dots
+#' @param time_step The time step of interest
 create_correlation <- function(norm_table, treatment, output_f, output_file,
-                               color_col = "stable") {
+                               color_col = "stable", time_step = 5) {
     dir.create(output_f, showWarnings = F)
     if (color_col == "stable") {
         my_colors <- c("dimgrey", "red")
@@ -127,15 +130,16 @@ create_correlation <- function(norm_table, treatment, output_f, output_file,
         )
     }
     sg <- sum(norm_table$stable)
+    coly <- paste0("T_", time_step)
     p <- ggplot(norm_table, mapping = aes(
-        x = log10(`T_0`), y = log10(`T_5`),
+        x = log10(`T_0`), y = log10(!!as.symbol(coly)),
         color = !!as.symbol(color_col)
     )) +
         scale_color_manual(values = my_colors) +
         geom_abline(slope = 1, color = "blue", size = 2) +
         geom_point() +
         ggtitle(paste0(
-            "Correlation between 0h vs 5h of cells",
+            "Correlation between 0h vs", time_step, "h of cells",
             "treated with triptolite and ", treatment, " (", sg,
             " stable genes)"
         ))
@@ -157,7 +161,6 @@ create_correlation <- function(norm_table, treatment, output_f, output_file,
 #' @param colors The colors of groups of genes
 create_expression_boxplot <- function(norm_table, treatment, output_f,
                                       output_file, my_colors) {
-    message(my_colors)
     new_table <- norm_table %>%
         pivot_longer(c(-gene, -stable, -group),
             names_to = "condition",
@@ -188,9 +191,11 @@ create_expression_boxplot <- function(norm_table, treatment, output_f,
 #' @param output_file File that will contain the correlation figure
 #' @param stable_file A file containing stable genes
 #' @param output_f The folder were the results will be created
+#' @param time_step The time step of interest
 #' @import tidyverse
-get_mean_correleation_figure <- function(norm_matrix, output_file,
-                                         stable_file, treatment, output_f) {
+get_mean_correlation_figure <- function(norm_matrix, output_file,
+                                        stable_file, treatment, output_f,
+                                        time_step) {
     res <- norm_matrix %>%
         as_tibble() %>%
         pivot_longer(-gene, values_to = "counts", names_to = "sample") %>%
@@ -208,8 +213,14 @@ get_mean_correleation_figure <- function(norm_matrix, output_file,
         mutate(stable = gene %in% stable_gene) %>%
         arrange(stable) # nolint
     res <- get_group_columns(res) # nolint
-    create_correlation(res, treatment, output_f, output_file, "stable")
-    create_correlation(res, treatment, output_f, output_file, "group")
+    create_correlation(
+        res, treatment, output_f, output_file, "stable",
+        time_step
+    )
+    create_correlation(
+        res, treatment, output_f, output_file, "group",
+        time_step
+    )
 }
 
 
@@ -228,13 +239,16 @@ normalize_on_stable_gene <- function(count_threshold = 5e6,
                                          "results/deseq2_normalisation") {
     my_pattern <- paste0(
         treatment, "_T*_0|",
-        treatment, "_T*_5|", treatment, "_DMSO_0|",
-        treatment, "_TCHX_5"
+        treatment, "_T*_5|",
+        treatment, "_DMSO_0|",
+        treatment, "_TCHX_5|",
+        treatment, "_T*_3|",
+        treatment, "_TCHX_3"
     )
     full_data <- load_n_filter_data(
         sample_threshold = count_threshold,
         pattern = my_pattern,
-        gene_threshold = 0,
+        gene_threshold = 10,
         load_ercc = use_ercc
     )
     if (treatment == "BRAF") {
@@ -255,12 +269,14 @@ normalize_on_stable_gene <- function(count_threshold = 5e6,
     if (use_ercc) {
         name_ercc <- "_ercc"
     }
-    fig_name <- paste0(treatment, name_ercc, "_T0_vs_T5_mean")
+    for (ts in c("3", "5")) {
+        fig_name <- paste0(treatment, name_ercc, "_T0_vs_T", ts, "_mean")
         norm_table <- data.frame(gene = rownames(norm_counts), norm_counts)
-    get_mean_correleation_figure(
+        get_mean_correlation_figure(
             norm_table,
             fig_name, stable_file, treatment,
-        output_f = output_folder
+            output_f = output_folder,
+            time_step = ts
         )
         write_tsv(norm_table,
             file = paste0(
@@ -269,6 +285,7 @@ normalize_on_stable_gene <- function(count_threshold = 5e6,
             )
         )
     }
+}
 
 
 #' Normalise every conditions separatly
@@ -276,20 +293,20 @@ normalize_on_stable_gene <- function(count_threshold = 5e6,
 #' @param output_folder Folder where the results will be created
 grep_normalise_conditions <- function(output_folder = "results/deseq2_normalisation") {
     normalize_on_stable_gene(
-        count_threshold = 5e6, treatment = "DMSO",
+        count_threshold = 3e6, treatment = "DMSO",
         output_folder = output_folder
     )
     normalize_on_stable_gene(
-        count_threshold = 4.8e6, treatment = "BRAF",
+        count_threshold = 3e6, treatment = "BRAF",
         output_folder = output_folder
     )
     normalize_on_stable_gene(
-        count_threshold = 5e6, treatment = "DMSO",
+        count_threshold = 3e6, treatment = "DMSO",
         use_ercc = T,
         output_folder = output_folder
     )
     normalize_on_stable_gene(
-        count_threshold = 4.8e6, treatment = "BRAF",
+        count_threshold = 3e6, treatment = "BRAF",
         use_ercc = T,
         output_folder = output_folder
     )