A human pancreas Visium spatial example dataset

A compact gene-filtered version of a human pancreatic intraepithelial neoplasia (PanIN) 10x Visium dataset from GSE254829. The object keeps the 1986 non-background tissue spots from sample GSM8058244 (PanIN-LG2), with a Spatial assay, a slice1 Visium image, and tissue coordinates in metadata columns x and y. Metadata column coda_label stores the dominant CODA microanatomical component for each spot, and coda_score stores its percentage. Component percentage columns are stored with the coda_ prefix, and the matched CODA table is stored in @tools$GSE254829_coda_table. To keep the package data small and directly usable with the bundled panc8_sub reference, the object retains the top 5000 genes shared with panc8_sub, ranked by total spatial counts.

Format

A Seurat object with 5000 genes, 1986 spots, and one Visium image named slice1.

Source

Derived from the GSE254829 human PanIN 10x Visium dataset: GSE254829. The package object uses the GEO supplementary files GSM8058244_PanIN-LG2.tar.gz and GSE254829_codatable_may202024.csv.gz.

Examples

if (interactive()) {
  download.file(

  "https://ftp.ncbi.nlm.nih.gov/geo/samples/GSM8058nnn/GSM8058244/suppl/GSM8058244_PanIN-LG2.tar.gz",
    destfile = "GSM8058244_PanIN-LG2.tar.gz",
    mode = "wb"
  )
  untar("GSM8058244_PanIN-LG2.tar.gz", exdir = ".")
  download.file(

  "https://ftp.ncbi.nlm.nih.gov/geo/series/GSE254nnn/GSE254829/suppl/GSE254829_codatable_may202024.csv.gz",
    destfile = "GSE254829_codatable_may202024.csv.gz",
    mode = "wb"
  )

  spatial <- Seurat::Load10X_Spatial(
    data.dir = "GSM8058244_PanIN-LG2",
    filename = "filtered_feature_bc_matrix"
  )
  tissue_pos <- utils::read.csv(
    file.path("GSM8058244_PanIN-LG2", "spatial", "tissue_positions_list.csv"),
    header = FALSE,
    row.names = 1
  )
  spatial <- subset(
    spatial,
    cells = intersect(colnames(spatial), rownames(tissue_pos)[tissue_pos$V2 == 1])
  )
  coords <- SeuratObject::GetTissueCoordinates(spatial)
  spatial$x <- coords[colnames(spatial), "x"]
  spatial$y <- coords[colnames(spatial), "y"]
  spatial$sample_id <- "GSM8058244"
  spatial$geo_accession <- "GSE254829"
  spatial$patient_sample <- "PanIN-LG2"

  coda <- utils::read.csv("GSE254829_codatable_may202024.csv.gz", check.names = FALSE)
  coda$barcode <- sub("_[0-9]+$", "", coda[[1]])
  coda <- coda[coda$sample == "PanIN-LG2" & coda$barcode %in% colnames(spatial), ]
  rownames(coda) <- coda$barcode
  coda <- coda[colnames(spatial), ]
comp_cols <- c("islets", "normal epithelium", "smooth muscle", "fat", "acini", "collagen",
  "panin")
  coda$coda_label <- names(coda[comp_cols])[
    max.col(as.matrix(coda[comp_cols]), ties.method = "first")
  ]
  coda$coda_score <- do.call(pmax, coda[comp_cols])
  spatial$coda_label <- coda$coda_label
  spatial$coda_score <- coda$coda_score
  spatial$panin_grade <- coda$paningrade
  for (cc in comp_cols) {
    spatial[[paste0("coda_", gsub(" ", ".", cc))]] <- coda[[cc]]
  }
  spatial@tools$GSE254829_coda_table <- coda[
    ,
    c(comp_cols, "nontissue", "paningrade", "coda_label", "coda_score")
  ]

  data(panc8_sub)
  shared <- intersect(rownames(spatial), rownames(panc8_sub))
  gene_counts <- GetAssayData5(spatial, assay = "Spatial", layer = "counts")
  genes_keep <- names(
    sort(Matrix::rowSums(gene_counts[shared, , drop = FALSE]), decreasing = TRUE)
  )[seq_len(5000)]
  spatial <- spatial[genes_keep, ]
  spatial$feature_selection <- "top_5000_shared_with_panc8_by_spatial_counts"

  use_data <- thisutils::get_namespace_fun("usethis", "use_data")
  use_data(
    spatial,
    compress = "xz",
    overwrite = TRUE
  )
}