Upload pull_gse236892.R with huggingface_hub

pull_gse236892.R

@@ -0,0 +1,138 @@
+library(GEOquery)
+library(SummarizedExperiment)
+library(tidyverse)
+library(here)
+library(AnnotationDbi)
+library(org.Hs.eg.db)
+library(GenomicRanges)
+library(txdbmaker)
+library(org.Hs.eg.db)
+
+# retrieve annotation info
+txdb_path <- here("~/projects/earli/data/txdb_gencode_v49.rds")
+
+if (file.exists(txdb_path)) {
+    txdb <- AnnotationDbi::loadDb(txdb_path)
+} else {
+    gtf_url <- "https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_49/gencode.v49.annotation.gtf.gz"
+    gtf_tmp <- tempfile(fileext = ".gtf.gz")
+    curl::curl_download(gtf_url, gtf_tmp, quiet = FALSE)
+    txdb <- txdbmaker::makeTxDbFromGFF(gtf_tmp, format = "gtf")
+    AnnotationDbi::saveDb(txdb, txdb_path)
+}
+
+hemoglobin_genes <- c(
+    ENSG00000206172 = "HBA1",
+    ENSG00000188536 = "HBA2",
+    ENSG00000244734 = "HBB",
+    ENSG00000229988 = "HBBP1",
+    ENSG00000223609 = "HBD",
+    ENSG00000213931 = "HBE1",
+    ENSG00000213934 = "HBG1",
+    ENSG00000196565 = "HBG2",
+    ENSG00000206177 = "HBM",
+    ENSG00000086506 = "HBQ1",
+    ENSG00000130656 = "HBZ",
+    ENSG00000206178 = "HBZP1"
+)
+
+# retrieve phenotype data
+gset <- getGEO("GSE236892", GSEMatrix = TRUE, getGPL = TRUE)
+gset <- gset[[1]]
+
+pdata <- pData(gset) |>
+    as_tibble(rownames = "geo_accession_row") |>
+    mutate(
+        sepsis_group  = str_extract(title, "(?<=_)(Hyper|Hypo)$"),
+        cnts_colnames = str_remove(title, "_(?<=_)(Hyper|Hypo)$")
+    )
+
+GSE236892_meta <- pdata |>
+    dplyr::select(geo_accession, cnts_colnames, `age:ch1`, `gender:ch1`, `lca:ch1`) |>
+    dplyr::rename(age = `age:ch1`, sex = `gender:ch1`, lca_label = `lca:ch1`) |>
+    column_to_rownames("cnts_colnames") |>
+    mutate(
+        lca_label   = factor(lca_label, levels = c("Hypo", "Hyper")),
+        sex         = factor(sex),
+        # label columns which had non-numeric age
+        imputed_age = age == "90+",
+        # strip non-numeric characters and cast to integer
+        age         = as.integer(str_remove(age, "\\+")),
+        # scale age to avoid near collinearity with the intercept. Note
+        # that this is only necessary or useful if age is used as a predictor
+        age_scaled  = as.numeric(scale(age))
+    ) |>
+    as_tibble(rownames = "sample_id")
+
+# retrieve count matrix
+supp_files <- getGEOSuppFiles("GSE236892", baseDir = tempdir(), fetch_files = TRUE)
+cnt_path <- rownames(supp_files)[str_detect(rownames(supp_files), "cnt_data")]
+
+cnt_mat <- read_csv(cnt_path) |>
+    dplyr::rename(ensg = `...1`) |>
+    column_to_rownames("ensg") |>
+    dplyr::select(all_of(GSE236892_meta$cnts_colnames)) |>
+    as.matrix()
+storage.mode(cnt_mat) <- "integer"
+
+
+gencode_genes <- suppressMessages(genes(txdb))
+names(gencode_genes) <- str_remove(names(gencode_genes), "\\.\\d+$")
+
+ensembl_meta <- AnnotationDbi::select(
+    org.Hs.eg.db,
+    keys    = names(gencode_genes),
+    columns = c("ENSEMBL", "SYMBOL", "GENENAME", "GENETYPE"),
+    keytype = "ENSEMBL"
+) |>
+    as_tibble() |>
+    filter(!is.na(ENSEMBL)) |>
+    distinct(ENSEMBL, .keep_all = TRUE)
+
+mcols(gencode_genes) <- ensembl_meta[
+    match(names(gencode_genes), ensembl_meta$ENSEMBL),
+    c("ENSEMBL", "SYMBOL", "GENENAME", "GENETYPE")
+]
+
+cnt_ensg <- rownames(cnt_mat)
+mapped <- gencode_genes[names(gencode_genes) %in% cnt_ensg]
+unmapped_ids <- setdiff(cnt_ensg, names(gencode_genes))
+
+unmapped <- GRanges(
+    seqnames = rep("chrUn", length(unmapped_ids)),
+    ranges   = IRanges(start = seq_along(unmapped_ids), width = 1),
+    ENSEMBL  = unmapped_ids,
+    SYMBOL   = NA_character_,
+    GENENAME = NA_character_,
+    GENETYPE = NA_character_
+)
+names(unmapped) <- unmapped_ids
+
+row_gr <- c(mapped, unmapped)[cnt_ensg]
+
+mcols(row_gr)$autosomal_protein_coding <- (
+    !is.na(mcols(row_gr)$GENETYPE) &
+        mcols(row_gr)$GENETYPE == "protein-coding" &
+        as.character(seqnames(row_gr)) %in% paste0("chr", 1:22)
+)
+
+mcols(row_gr)$hemoglobin_related <- names(row_gr) %in% names(hemoglobin_genes)
+
+feature_meta <- feature_meta <- as_tibble(row_gr) |>
+    dplyr::rename(feature_id = ENSEMBL) |>
+    dplyr::relocate(feature_id) |>
+    janitor::clean_names()
+
+stopifnot(identical(colnames(cnt_mat), GSE236892_meta$sample_id))
+stopifnot(identical(rownames(cnt_mat), feature_meta$feature_id))
+
+GSE236892_expr_long <- cnt_mat |>
+    as_tibble(rownames = "feature_id") |>
+    pivot_longer(-feature_id, names_to = "sample_id", values_to = "value") |>
+    arrange(sample_id, feature_id)
+
+dir.create("~/projects/hf_sepsis_collection/GSE236892")
+
+arrow::write_parquet(GSE236892_expr_long, "~/projects/hf_sepsis_collection/GSE236892/expression.parquet")
+arrow::write_parquet(feature_meta, "~/projects/hf_sepsis_collection/GSE236892/feature_metadata.parquet")
+arrow::write_parquet(GSE236892_meta, "~/projects/hf_sepsis_collection/GSE236892/sample_metadata.parquet")