library(MicroBioMap)
library(phyloseq)
library(mia)

if (!(file.exists("data/hmc-phyloseq-europe.Rds") || file.exists("data/hmc-phyloseq.Rds"))) {
    cpd <- getCompendium()
    cpd_phyloseq <- convertToPhyloseq(x = cpd)
    # sample_data(cpd_phyloseq)$geo_loc_name |> grepl(pattern = "europe")
    rm(cpd)
    gc()
    # sample_data(cpd_phyloseq)
    saveRDS(cpd_phyloseq, "data/hmc-phyloseq.Rds")
    cpd_phyloseq_europe <- subset_samples(cpd_phyloseq, grepl("europe", geo_loc_name))
    rm(cpd_phyloseq)
    gc()
    saveRDS(cpd_phyloseq_europe, "data/hmc-phyloseq-europe.Rds")
} else {
    cpd_phyloseq_europe <- readRDS("data/hmc-phyloseq-europe.Rds")
    cpd_phyloseq <- readRDS("data/hmc-phyloseq.Rds")
}