if ("MicroBioMap" %in% rownames(installed.packages()) == FALSE) {
    cat("MicroBioMap package not found. Installing...\n")
    if (!requireNamespace("BiocManager", quietly = TRUE)) {
        install.packages("BiocManager")
    }
    BiocManager::install("blekhmanlab/MicroBioMap")
}
library(MicroBioMap)
library(countrycode)
library(igraph)
library(here)
library(readr)
setClassUnion("ExpData", c("matrix", "SummarizedExperiment"))
cpd <- getCompendium()

colData(cpd)$country <- countrycode(colData(cpd)$iso, "iso2c", "country.name")

projects <- unique(colData(cpd)$project)

save_folder <- here("data")
if (!dir.exists(save_folder)) {
    dir.create(save_folder)
}

for (project in projects) {
    cat("Processing project:", project, "\n")
    # Filter the compendium for the current project
    project_data <- cpd[, colData(cpd)$project == project]
    matrix_data <- counts(project_data)
    # Create a graph from the matrix data
    g <- graph_from_biadjacency_matrix(matrix_data)
    edge_list <- as_edgelist(g) |> as.data.frame()
    colnames(edge_list) <- c("source", "target")
    edge_list_path <- here(save_folder, paste0(project, "_edge_list.csv.gz"))
    write_csv(edge_list, file = edge_list_path)
    supinfo_path <- here(save_folder, paste0(project, "_supinfo.csv.gz"))
    supinfo_df <- as.data.frame(colData(project_data))
    write_csv(supinfo_df, file = supinfo_path)
}

list_matrices <- lapply(projects, function(project) {
    project_data <- cpd[, colData(cpd)$project == project]
    matrix_data <- counts(project_data)
    matrix_data
})
names(list_matrices) <- projects

projects_supinfo <- colData(cpd)
projects_supinfo <- as.data.frame(projects_supinfo)

write_csv(projects_supinfo, file = here(save_folder, "projects_supinfo.csv.gz"))
saveRDS(list_matrices, file = here(save_folder, "list_matrices.rds"))