maj

Module_correlation.Rmd

@@ -4,6 +4,22 @@ output: html_document
 ---
 
 
+```{r input_parameters, include=FALSE}
+# Chemin  vers le fichier de design
+design_file <- "C:/Users/User/Desktop/projet_visualisation/design_WS3.csv"
+
+# Liste des noms de variables à considérer comme catégorielles.
+categorical_vars <- c("sample", "condition", "animal", "experiment", "extraction")
+
+# Liste des noms de variables à considérer comme quantitatives.
+quantitative_vars <- c( "volume", "quantity")
+
+# Liste des noms de variables à afficher pour la figure "matrice de plots" (advanced pairs plot).
+# Ici, on exclut par exemple 'sample' pour ne pas surcharger la visualisation.
+display_vars <- c("condition", "animal", "experiment", "extraction")
+```
+
+
 ```{r setup, include=FALSE}
 
 # Global options: hide code, warnings and messages in the final report.
@@ -27,6 +43,10 @@ if(!require(ComplexHeatmap)) {
 }
 if(!require(RColorBrewer)) install.packages("RColorBrewer")
 if(!require(reshape2)) install.packages("reshape2")
+if(!require(DT)) install.packages("DT")
+if(!require(plotly)) install.packages("plotly")
+if(!require(lsr)) install.packages("lsr")
+if(!require(knitr)) install.packages("knitr")
 
 library(GGally)
 library(readr)
@@ -36,6 +56,13 @@ library(corrplot)
 library(ComplexHeatmap)
 library(RColorBrewer)
 library(reshape2)
+library(DT)
+library(plotly)
+library(lsr)  
+library(knitr)
+
+
+
 
 
 ```
@@ -78,31 +105,23 @@ cat("Variables catégorielles détectées : ", paste(colnames(cat_data), collaps
 ```
 
 2.1 Automated Conversion for Correlation Analysis
-Note: For correlation analysis, only the quantitative variables are used.
+
 ```{r}
 data_numeric <- annotations %>%
   mutate(
-    condition = as.numeric(factor(condition)),  # Conversion des catégories en indices numériques
-    animal = as.numeric(factor(animal)),
-    experiment = as.numeric(as.Date(experiment, format = "%d.%m.%Y")),  # Conversion en nombre
-    extraction = as.numeric(as.Date(extraction, format = "%d.%m.%Y")),
-    sample = as.numeric(factor(sample))  # Conversion de 'sample' en indice numérique
-  ) %>%
-  select(-sample)  # Exclure 'sample' du calcul de corrélation
-
-str(data_numeric)
-
+    condition = as.numeric(factor(condition)),  # Conversion de 'condition'
+    animal = as.numeric(factor(animal))           # Conversion de 'animal'
 
+    # 'sample' sera exclue du calcul
+  ) %>%
+  select(-sample, -experiment, -extraction)
 ```
 3. Visualizations of Correlations and Distributions
 3.1 Correlation Matrix with corrplot
 Figure: Correlation matrix showing pairwise Pearson correlations among quantitative variables.
 ```{r}
 if(ncol(data_numeric) > 1){
-  # Calcul de la matrice de corrélation (méthode Pearson)
   cor_matrix <- cor(data_numeric, use = "complete.obs", method = "pearson")
-  
-  # Génération de la figure de corrélation
   corrplot(cor_matrix, method = "circle", type = "lower",
            tl.col = "black", tl.cex = 0.8,
            col = colorRampPalette(c("blue", "white", "red"))(200))
@@ -111,6 +130,7 @@ if(ncol(data_numeric) > 1){
 }
 
 
+
 ```
 
 3.2 Heatmap with ComplexHeatmap
@@ -143,8 +163,13 @@ if(ncol(num_data) > 1){
 
 3.4 Distribution of Numeric Variables
 Figure: Histograms with density curves for each quantitative variable.
-```{r}
-numeric_vars <- names(data_numeric)
+
+```{r distribution_numeric}
+
+# Sélectionner uniquement les variables quantitatives qui existent dans data_numeric
+numeric_vars <- quantitative_vars[quantitative_vars %in% names(data_numeric)]
+
+# Génération d'un histogramme et d'une courbe de densité pour chaque variable quantitative
 for (var in numeric_vars) {
   p <- ggplot(data_numeric, aes_string(x = var)) +
     geom_histogram(aes(y = ..density..), bins = 30, fill = "blue", alpha = 0.5) +
@@ -156,6 +181,7 @@ for (var in numeric_vars) {
 }
 
 
+
 ```
 
 3.5 Scatter Plots for Highly Correlated Pairs
@@ -188,63 +214,80 @@ if(ncol(data_numeric) > 1){
 ```
 
 3.6 Distribution of Categorical Variables
-Figure: Frequency distributions (bar charts) for each categorical variable.
 
+## Frequency Distributions for Categorical Variables {.tabset}
+
+
+### experiment
 ```{r}
-cat_vars <- names(cat_data)
-for (var in cat_vars) {
-  p <- ggplot(annotations, aes_string(x = var)) +
-       geom_bar(fill = "blue", alpha = 0.7) +
-       labs(title = paste("Frequency Distribution of", var),
-            x = var, y = "Count") +
-       theme_minimal()
-  print(p)
-}
 
+ggplot(annotations, aes(x = experiment)) +
+  geom_bar(fill = "purple", alpha = 0.7) +
+  labs(title = "Distribution: experiment", x = "Experiment", y = "Count") +
+  theme_minimal()
+```
+
+### extraction
+```{r}
+
+ggplot(annotations, aes(x = extraction)) +
+  geom_bar(fill = "orange", alpha = 0.7) +
+  labs(title = "Distribution: extraction", x = "Extraction", y = "Count") +
+  theme_minimal()
+```
+
+
+3.7 Sample Visualizations
+
+ 3.7.1 Number of Samples by Condition
+
+```{r sample_barplot, echo=FALSE, fig.height=5, fig.width=7}
+# Bar plot: Nombre d'échantillons par condition
+
+p1 <- ggplot(annotations, aes(x = condition)) +
+  geom_bar(fill = "steelblue", alpha = 0.7) +
+  labs(title = "Number of Samples by Condition",
+       x = "Condition", y = "Count") +
+  theme_minimal()
+print(p1)
 
 ```
-3.7 Detailed Visualization of the 'sample' Variable
-Even though the unique sample identifiers are not necessarily used in every figure, it is important to retain and examine their distribution.
-Here we provide a dedicated bar chart for the sample variable.
 
+# Table interactive: Liste des identifiants d'échantillons associées aux  variables condition et animal
 ```{r}
-if("sample" %in% names(annotations)){
-  # Ensure 'sample' is treated as a factor and preserve its order.
-  annotations$sample <- factor(annotations$sample, levels = unique(annotations$sample))
-  
-  # Bar chart for 'sample'
-  p <- ggplot(annotations, aes(x = sample)) +
-       geom_bar(fill = "steelblue", alpha = 0.7) +
-       labs(title = "Frequency Distribution of 'sample'",
-            x = "Sample ID", y = "Count") +
-       theme_minimal()
-  print(p)
-} else {
-  cat("The variable 'sample' does not exist in the dataset.\n")
-}
+datatable(annotations[, c("sample", "condition", "animal")],
+          options = list(pageLength = 10),
+          caption = "Table of Sample Identifiers and Associated Groups")
 
 ```
 
+
+
+
 3.8 Advanced Visualization: Improved Pairs Plot
 Figure: Advanced pairs plot of selected categorical variables (excluding 'sample') to provide a global view of group relationships.
 
 ```{r}
-# Create a copy of the data for the pairs plot.
-# Retain 'sample' in the dataset but remove it from the pairs plot display.
-data_factors <- annotations %>% select(any_of(c(display_vars, "sample")))
-data_pairs <- data_factors %>% select(-sample)
-
-# Ensure that the variables to be displayed are factors.
-for (col in display_vars) {
-  if(col %in% names(data_pairs)){
-    data_pairs[[col]] <- as.factor(data_pairs[[col]])
+# Créer une copie des données pour conversion en facteurs
+data_factors <- annotations
+
+# Conversion des colonnes spécifiques en facteurs si elles existent
+cols_to_factor <- c("condition", "animal", "experiment", "extraction")
+for (col in cols_to_factor) {
+  if (col %in% names(data_factors)) {
+    data_factors[[col]] <- as.factor(data_factors[[col]])
   }
 }
 
-# Define custom colors for the 'condition' variable.
-custom_colors <- c("#1B9E77", "#D95F02", "#7570B3")
+# Supprimer la colonne "sample" pour cette visualisation
+if("sample" %in% names(data_factors)) {
+  data_factors <- data_factors %>% select(-sample)
+}
+
+# Définition des couleurs personnalisées
+custom_colors <- c("#1B9E77", "#D95F02", "#7570B3")  # Palette pour 'condition'
 
-advanced_plot <- ggpairs(data_pairs, 
+ggpairs(data_factors, 
         mapping = aes(color = condition),
         lower = list(
           continuous = wrap("points", alpha = 0.7, size = 2),
@@ -254,72 +297,150 @@ advanced_plot <- ggpairs(data_pairs,
           continuous = wrap("densityDiag", alpha = 0.5, fill = custom_colors[2])
         ),
         upper = list(
-          continuous = wrap("cor", size = 4, color = "black")
+          continuous = wrap("cor", size = 5, color = "black")
         )
 ) +
-  scale_color_manual(values = custom_colors) +
   theme_bw() +
   theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 10),
         axis.text.y = element_text(size = 10),
-        legend.position = "bottom") +
-  ggtitle("Advanced Pairs Plot of Selected Variables")
-advanced_plot
+        legend.position = "bottom")
+
 
 
 ```
 
-4. Frequency Distributions of Categorical Variables (Tabset)
-The following section uses R Markdown tabsets to display the frequency histograms of each categorical variable in separate tabs.
-This helps in quickly verifying that each level of a variable is represented by a similar number of samples—a key point for balanced downstream analyses.
+## Group Balance Check
 
-<!-- To enable tabbed sections, add {.tabset} to the header -->
-Frequency Distributions for Categorical Variables {.tabset}
-condition
 
+ Vérifier la répartition des échantillons par condition
 ```{r}
-ggplot(annotations, aes(x = condition)) +
-  geom_bar(fill = "blue", alpha = 0.7) +
-  labs(title = "Frequency Distribution: condition",
-       x = "Condition", y = "Count") +
-  theme_minimal()
+condition_counts <- annotations %>%
+  count(condition) %>%
+  arrange(desc(n))
+print(condition_counts)
 
 ```
-
-sample
+ 
+ Vérifier la répartition des échantillons par animal
 ```{r}
-ggplot(annotations, aes(x = sample)) +
-  geom_bar(fill = "steelblue", alpha = 0.7) +
-  labs(title = "Frequency Distribution: sample",
-       x = "Sample ID", y = "Count") +
-  theme_minimal()
-
+animal_counts <- annotations %>%
+  count(animal) %>%
+  arrange(desc(n))
+print(animal_counts)
 ```
-animal
+ 
+ Bar plot interactif pour la répartition par animal
 ```{r}
-ggplot(annotations, aes(x = animal)) +
-  geom_bar(fill = "blue", alpha = 0.7) +
-  labs(title = "Frequency Distribution: animal",
-       x = "Animal", y = "Count") +
+p_animal <- ggplot(animal_counts, aes(x = reorder(animal, -n), y = n)) +
+  geom_bar(stat = "identity", fill = "darkgreen", alpha = 0.8) +
+  labs(title = "Number of Samples by Animal",
+       x = "Animal",
+       y = "Count") +
   theme_minimal()
-
+ggplotly(p_animal)
 ```
-
-experiment
+ 
+ Sous-échantillonnage pour équilibrer les groupes par condition
 ```{r}
-ggplot(annotations, aes(x = experiment)) +
-  geom_bar(fill = "blue", alpha = 0.7) +
-  labs(title = "Frequency Distribution: experiment",
-       x = "Experiment", y = "Count") +
-  theme_minimal()
+min_count <- min(condition_counts$n)
+balanced_annotations <- annotations %>%
+  group_by(condition) %>%
+  sample_n(min_count) %>%
+  ungroup()
 
+cat("Subsampled to", min_count, "samples per condition.\n")
 ```
-extraction
-```{r}
-ggplot(annotations, aes(x = extraction)) +
-  geom_bar(fill = "blue", alpha = 0.7) +
-  labs(title = "Frequency Distribution: extraction",
-       x = "Extraction", y = "Count") +
-  theme_minimal()
 
-```
 
+
+## 4. Association Matrix: Matrice des liaisons
+
+```{r association_matrix, echo=FALSE, fig.width=8, fig.height=6}
+# --- Définition des fonctions d'association ---
+
+# 1. Fonction pour calculer le coefficient de corrélation ratio (η) entre une variable numérique et une variable catégorielle
+correlationRatio <- function(x, group) {
+  df <- data.frame(x, group)
+  df <- df[complete.cases(df), ]
+  x <- df$x
+  group <- df$group
+  grandMean <- mean(x)
+  # Somme des carrés entre les groupes
+  ssBetween <- tapply(x, group, function(xg) length(xg) * (mean(xg) - grandMean)^2)
+  ssBetween <- sum(ssBetween)
+  # Somme des carrés dans les groupes
+  ssWithin <- tapply(x, group, function(xg) sum((xg - mean(xg))^2))
+  ssWithin <- sum(ssWithin)
+  eta2 <- ssBetween / (ssBetween + ssWithin)
+  return(sqrt(eta2))
+}
+
+# 2. Fonction pour calculer le V de Cramér à partir d'un tableau de contingence
+computeCramersV <- function(tab) {
+  # Calcul du chi2 sans correction de continuité
+  chi2 <- suppressWarnings(chisq.test(tab, correct = FALSE)$statistic)
+  n <- sum(tab)
+  phi2 <- chi2 / n
+  r <- nrow(tab)
+  c <- ncol(tab)
+  # Correction pour éviter des valeurs négatives
+  phi2corr <- max(0, phi2 - ((c - 1) * (r - 1)) / (n - 1))
+  rcorr <- r - ((r - 1)^2) / (n - 1)
+  ccorr <- c - ((c - 1)^2) / (n - 1)
+  return(sqrt(phi2corr / min((ccorr - 1), (rcorr - 1))))
+}
+
+# 3. Fonction pour calculer l'association entre deux variables :
+#    - Si les deux variables sont numériques : corrélation de Pearson.
+#    - Si les deux variables sont catégorielles : V de Cramér (via computeCramersV).
+#    - Si l'une est numérique et l'autre catégorielle : corrélation ratio (η).
+computeAssociation <- function(x, y) {
+  if (is.numeric(x) && is.numeric(y)) {
+    return(cor(x, y, use = "complete.obs"))
+  } else if (is.factor(x) && is.factor(y)) {
+    tab <- table(x, y)
+    return(computeCramersV(tab))
+  } else if (is.numeric(x) && is.factor(y)) {
+    return(correlationRatio(x, y))
+  } else if (is.factor(x) && is.numeric(y)) {
+    return(correlationRatio(y, x))
+  } else {
+    return(NA)
+  }
+}
+
+# 4. Fonction pour construire la matrice des liaisons pour toutes les variables d'un jeu de données
+associationMatrix <- function(data) {
+  vars <- names(data)
+  nvar <- length(vars)
+  M <- matrix(NA, nrow = nvar, ncol = nvar, dimnames = list(vars, vars))
+  for (i in seq_len(nvar)) {
+    for (j in seq_len(nvar)) {
+      M[i, j] <- computeAssociation(data[[i]], data[[j]])
+    }
+  }
+  return(M)
+}
+
+# --- Préparation du jeu de données pour la matrice d'association ---
+# On utilise les variables quantitatives (par exemple, "volume", "quantity") et 
+# les variables catégorielles définies dans display_vars (excluant "sample").
+mixed_vars <- c(quantitative_vars, setdiff(display_vars, "sample"))
+mixed_data <- annotations %>% select(all_of(mixed_vars))
+
+# S'assurer que les variables catégorielles sont bien de type factor
+for (var in display_vars) {
+  if (var %in% names(mixed_data)) {
+    mixed_data[[var]] <- as.factor(mixed_data[[var]])
+  }
+}
+
+# Calcul de la matrice des liaisons
+assoc_mat <- associationMatrix(mixed_data)
+
+# Affichage de la matrice sous forme de tableau
+knitr::kable(assoc_mat, digits = 2, caption = "Association Matrix (Matrice des liaisons)")
+
+
+
+```