为何无法修改SingleCellExperiment对象的行名与列名？求协助

Let's break down why your row and column names aren't behaving as expected, and fix the code step by step.

The Issue in Your Original Code #

Your current code doesn't properly assign row/column names to the SingleCellExperiment (SCE) object because:

• You're not setting the row names of your expression matrix (nsclc) to the gene IDs from the first column.
• The colData you created is just generic cell_1, cell_2... labels instead of linking to the actual cell metadata from meta_1, and there's no alignment between the expression matrix columns and metadata rows.
• Using a data.frame directly in metadata can cause structural issues, as SCE expects metadata to be a list.

Corrected Code #

Here's the revised script that properly sets up row/column names and aligns your data:

# Read expression matrix, set first column as gene row names directly
nsclc <- read.table(
  "D:\\Bioinformatica\\Project Work and PhD Proposal\\Data\\Data Extracted\\GSE143423_lbm_scRNAseq_gene_expression_counts.csv",
  header = TRUE,
  sep = ",",
  row.names = 1  # Critical: Assigns first column as gene IDs for the matrix
)

# Read metadata and clean up
meta_1 <- read.table(
  "D:\\Bioinformatica\\Project Work and PhD Proposal\\Data\\Data Extracted\\GSE143423_lbm_scRNAseq_metadata.csv",
  header = TRUE,
  sep = ","
)
meta_1$X <- NULL

# Align metadata rows with expression matrix columns
# Assuming your expression matrix's column names are the cell identifiers
# If meta_1 has a column with cell IDs (e.g., named "cell_id"), use that instead:
# row.names(meta_1) <- meta_1$cell_id
row.names(meta_1) <- colnames(nsclc)

# Create properly formatted SCE object
sce_lc <- SingleCellExperiment(
  assays = list(counts = as.matrix(nsclc)),  # Counts matrix inherits gene row names and cell column names
  rowData = DataFrame(gene_id = rownames(nsclc)),  # Store gene info in rowData (uses SCE's recommended DataFrame)
  colData = meta_1,  # colData rows match counts columns (cell IDs)
  metadata = list(experiment_id = "GSE143423", study_type = "NSCLC scRNA-seq")  # Metadata stored as a list for flexibility
)

Key Fixes Explained #

1. Expression Matrix Row Names: Adding row.names=1 when reading nsclc ensures your gene IDs become the row names of the counts matrix, which the SCE object will automatically use as its row names.
2. Metadata Alignment: Setting row.names(meta_1) <- colnames(nsclc) guarantees that each row in meta_1 corresponds exactly to a column (cell) in your counts matrix—this is a requirement for valid SCE objects.
3. Proper Data Structures: Using DataFrame for rowData and a list for metadata follows SCE's design standards, preventing unexpected behavior with data types and access.

Verify the Fix #

Run these commands to confirm your row/column names are correctly assigned:

# Check gene row names
head(rownames(sce_lc))

# Check cell column names
head(colnames(sce_lc))

# Confirm metadata rows match cell column names
all(colnames(sce_lc) == rownames(colData(sce_lc)))

内容的提问来源于stack exchange，提问作者Spartan 117