From fccfe90b3cc1eeea3f10055998ad21e2f7c93f57 Mon Sep 17 00:00:00 2001
From: Louis Lacoste <louis.lacoste@hotmail.fr>
Date: Fri, 7 Jun 2024 15:44:43 +0200
Subject: [PATCH] Adding smaller clustering

---
 .../01_dore_no_small_clusterize.R             | 123 ++++++++++++++++++
 .../migale_application_dore_no_small.sh       |  34 +++++
 2 files changed, 157 insertions(+)
 create mode 100644 code/applications/dore_no_small/01_dore_no_small_clusterize.R
 create mode 100755 code/scripts/migale_application_dore_no_small.sh

diff --git a/code/applications/dore_no_small/01_dore_no_small_clusterize.R b/code/applications/dore_no_small/01_dore_no_small_clusterize.R
new file mode 100644
index 0000000..42b7814
--- /dev/null
+++ b/code/applications/dore_no_small/01_dore_no_small_clusterize.R
@@ -0,0 +1,123 @@
+library(colSBM)
+library(dplyr)
+library(tidyr)
+library(here)
+
+# Arguments
+arg <- commandArgs(trailingOnly = TRUE)
+
+model <- "iid"
+seed <- 1234L
+
+if (length(arg) == 0L) {
+    message("No arguments provided, using default.")
+} else {
+    if ("--model" %in% arg) {
+        model <- arg[(which(arg == "--model") + 1L)]
+    } else {
+        message("No model provided, defaulting to iid.")
+    }
+    if ("--seed" %in% arg) {
+        seed <- try(as.integer(arg[(which(arg == "--seed") + 1L)]))
+    } else {
+        message("No seed provided, defaulting to 1234.")
+    }
+}
+
+#  Arguments checks
+allowed_model <- c("iid", "pi", "rho", "pirho")
+stopifnot(
+    "Unknown model, should be : iid, pi, rho or pirho" = (model %in% allowed_model),
+    "Seed isn't castable to integer" = (is.integer(seed))
+)
+
+message(
+    "Début du clustering des données Doré et al. sans les petits réseaux avec le modèle ", model,
+    " et la seed ", seed
+)
+
+set.seed(seed, "L'Ecuyer-CMRG")
+base_data_folder <- file.path(here(), "code", "data", "dore")
+save_folder <- file.path(here(), "code", "results", "applications", "dore_no_small")
+
+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
+
+min_size <- c(25L, 50L)
+
+incidence_matrices <- incidence_matrices[
+    (colSums(sapply(incidence_matrices, dim) > min_size) == 2L)
+]
+
+start_time <- format(Sys.time(), "%d-%m-%y_%H-%M-%S")
+
+list_collection <- clusterize_bipartite_networks(
+    netlist = incidence_matrices,
+    net_id = names(incidence_matrices),
+    colsbm_model = model,
+    nb_run = 3L,
+    global_opts = list(
+        nb_cores = parallelly::availableCores(omit = 1L),
+        plot_details = 0L,
+        backend = "parallel"
+    )
+)
+
+save_file <- file.path(
+    save_folder, paste0(
+        "dore_collection_no_small_",
+        model, "_seed_", seed,
+        "_", start_time, ".Rds"
+    )
+)
+
+message("Clustering saved.")
+saveRDS(list(min_size = min_size, clustering = list_collection),
+    file = save_file
+)
diff --git a/code/scripts/migale_application_dore_no_small.sh b/code/scripts/migale_application_dore_no_small.sh
new file mode 100755
index 0000000..f117fc7
--- /dev/null
+++ b/code/scripts/migale_application_dore_no_small.sh
@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+#$ -V
+#$ -cwd
+#$ -N Dore_app
+#$ -m besa
+#$ -t 1:4
+#$ -q long.q
+#$ -pe thread 64
+#$ -M louis.lacoste+migale@agroparistech.fr
+#$ -o logs/$JOB_NAME.$TASK_ID
+#$ -e logs/$JOB_NAME.$TASK_ID
+
+# Creating log directory if it doesn't exists
+BASE_DIR="/home/$USER/work/mia-stage-2024"
+LOG_DIR=$(echo "$BASE_DIR/logs")
+
+if [ ! -d "$LOG_DIR" ]; then
+    mkdir -p $LOG_DIR
+fi
+
+# Constant data
+MODELARRAY=("iid" "pi" "rho" "pirho")
+ID=$((SGE_TASK_ID - 1))
+MODEL=${MODELARRAY[$((ID % 4))]}
+
+SEED=0
+
+
+# Finding directory
+APPLICATIONS_DIR=$(echo "$BASE_DIR/code/applications")
+
+echo $APPLICATIONS_DIR
+
+Rscript "${APPLICATIONS_DIR}/dore_no_small/01_dore_no_small_clusterize.R" --model $MODEL --seed $SEED