for report : adding baldock network generation

code/for_report/applications/baldock.R

@@ -0,0 +1,117 @@
+library("ggplot2")
+library("ggokabeito")
+library("tidyr")
+library("dplyr")
+library("stringr")
+library("knitr")
+library("kableExtra")
+library("stringr")
+library("here")
+library("tikzDevice")
+library("network")
+library("GGally")
+library("igraph")
+library("tidygraph")
+library("ggraph")
+
+base_data_folder <- file.path(here(), "code", "data", "dore")
+save_folder <- file.path(
+    here(), "mia-rapport-2024", "results", "applications",
+    "sub-dore", "baldock"
+)
+
+if (!dir.exists(save_folder)) {
+    dir.create(save_folder, recursive = TRUE)
+}
+
+collection_data <- file.path(
+    base_data_folder,
+    "dore-matrices.Rds"
+)
+
+interaction_data <- read.table(file = file.path(base_data_folder, "interaction-data.txt"), sep = "\t", header = TRUE)
+
+seq_ids_network_aggreg <- unique(interaction_data$id_network_aggreg)
+names_aggreg_networks <- sapply(
+    seq_ids_network_aggreg,
+    function(id) {
+        paste0(
+            unique(interaction_data[which(interaction_data$id_network_aggreg == id), ]$web),
+            collapse = "+"
+        )
+    }
+)
+# Computation of incidence matrices
+incidence_matrices <- lapply(
+    seq_ids_network_aggreg,
+    function(m) {
+        current_interaction_data <- interaction_data[which(interaction_data$id_network_aggreg == m), ] %>%
+            mutate(
+                plantaggreg = paste(plantorder,
+                    plantfamily, plantgenus, plantspecies,
+                    sep = "-"
+                ),
+                insectaggreg = paste(insectorder,
+                    insectfamily, insectgenus, insectspecies,
+                    sep = "-"
+                )
+            )
+        current_interaction_data <- table(current_interaction_data$plantaggreg, current_interaction_data$insectaggreg)
+
+        current_incidence_matrix <- matrix(current_interaction_data,
+            ncol = ncol(current_interaction_data), dimnames = dimnames(current_interaction_data)
+        )
+
+        current_incidence_matrix[which(current_incidence_matrix > 0)] <- 1
+        return(current_incidence_matrix)
+    }
+)
+
+names(incidence_matrices) <- names_aggreg_networks
+
+incidence_matrices <- incidence_matrices[grepl(
+    x = names(incidence_matrices),
+    pattern = "Baldock"
+)]
+
+oi_palette <- palette_okabe_ito(1:2)
+names(oi_palette) <- c("plant", "insect")
+
+set.seed(1234)
+
+inc_mat <- incidence_matrices[[1]]
+aaa <- incidence_matrices[[2]]
+
+plot_list <- lapply(seq_along(incidence_matrices), function(id) {
+    incidence_matrices[[id]] |>
+        as.matrix() |>
+        as_tbl_graph() |>
+        mutate(node_type = ifelse(type, "Insect", "Plant"), graph_id = id) |>
+        ggraph(layout = "igraph", algorithm = "bipartite") +
+        geom_node_point(aes(color = node_type)) +
+        geom_edge_link(alpha = 0.1) +
+        scale_color_okabe_ito(order = 6:7) +
+        theme_void() +
+        labs(color = "Node type") +
+        theme(legend.position = "bottom")
+})
+names(plot_list) <- names(incidence_matrices)
+
+output_tikz_folder <- here(
+    "mia-rapport-2024", "tikz", "applications",
+    "baldock"
+)
+if (!dir.exists(output_tikz_folder)) {
+    dir.create(output_tikz_folder, recursive = TRUE)
+}
+options(tikzDocumentDeclaration = "\\documentclass[10pt]{standalone}")
+lapply(seq_along(plot_list), function(id) {
+    tikz(
+        file = file.path(output_tikz_folder, paste0("graph-", names(plot_list)[[id]], ".tex")),
+        width = 3,
+        height = 1,
+        standAlone = TRUE
+    )
+    print(plot_list[[id]])
+    dev.off()
+})