GSE13067 Processing Pipeline

Publication

Life science alliance (2020) — PMID 32817263

Dataset

Expression data from primary colorectal cancers

1. 1

   The simpleaffy package in R/Bioconductor was used to calculate MAS5.0 calls.

   R vInferred with models/gemini-2.5-flash

   $ Bash example

   # Install R and Bioconductor (if not already installed)
   # For example, using conda:
   # conda create -n r_simpleaffy r-base bioconductor-simpleaffy bioconductor-affy -y
   # conda activate r_simpleaffy
   
   # Or, within R:
   # if (!requireNamespace("BiocManager", quietly = TRUE))
   #     install.packages("BiocManager")
   # BiocManager::install(c("simpleaffy", "affy"))
   
   # Create an R script to calculate MAS5.0 calls
   cat << 'EOF' > calculate_mas5_calls.R
   # Load necessary libraries
   library(simpleaffy)
   library(affy) # Often used to read CEL files
   
   # Define the path to your CEL files
   # Replace with the actual path to your Affymetrix .CEL files
   cel_file_directory <- "." # Example: current directory, or specify e.g., "/data/affymetrix_cels"
   
   # List all .CEL files in the specified directory
   # The pattern uses double backslashes for escaping the dot in R regex
   cel_files <- list.files(path = cel_file_directory, pattern = ".*\\.CEL$", full.names = TRUE)
   
   if (length(cel_files) == 0) {
       stop("No .CEL files found in the specified directory: ", cel_file_directory, ". Please provide Affymetrix .CEL files.")
   }
   
   # Read the .CEL files into an AffyBatch object
   # This step requires the 'affy' package
   raw_data <- ReadAffy(filenames = cel_files)
   
   # Calculate MAS5.0 calls
   # This function returns a data frame with probe set IDs, detection p-values, and call (P/M/A)
   mas5_calls_results <- mas5calls(raw_data)
   
   # Define the output file path
   output_file <- "mas5_calls_output.csv"
   
   # Save the results to a CSV file
   write.csv(mas5_calls_results, output_file, row.names = FALSE)
   
   message(paste("MAS5.0 calls successfully calculated and saved to:", output_file))
   EOF
   
   Rscript calculate_mas5_calls.R

   Copy
2. 2

   These values are not normalized.

   Raw Data (Inferred with models/gemini-2.5-flash) vN/A

   $ Bash example

   # This step indicates that the data values are in their raw, unnormalized state.
   # No specific command is executed for this descriptive step itself.
   # Normalization would typically be performed by a subsequent tool.

   Copy
3. 3

   Normalization was performed by quantile normalization with respect to a defined reference set.

   R (preprocessCore) (Inferred with models/gemini-2.5-flash) vR 4.3.2, preprocessCore 1.64.0

   $ Bash example

   # Install R and preprocessCore package if not already installed
   # conda install -c conda-forge r-base=4.3.2
   # conda install -c bioconda r-preprocesscore=1.64.0
   
   # Create dummy input files for demonstration (replace with actual paths and data)
   # input_data.tsv: Tab-separated matrix with gene/feature names as row names and samples as columns
   # reference_distribution.tsv: Single-column file containing the sorted values of the target distribution
   
   # Example dummy input_data.tsv (replace with your actual data file)
   # echo -e "gene\tsample1\tsample2\tsample3" > input_data.tsv
   # echo -e "geneA\t100\t120\t90" >> input_data.tsv
   # echo -e "geneB\t50\t60\t45" >> input_data.tsv
   # echo -e "geneC\t200\t210\t180" >> input_data.tsv
   # echo -e "geneD\t10\t15\t8" >> input_data.tsv
   
   # Example dummy reference_distribution.tsv (replace with your actual reference distribution file)
   # This file should contain a single column of sorted values representing the target distribution.
   # echo -e "10\n50\n100\n200" > reference_distribution.tsv
   
   # R script for quantile normalization with a defined reference set
   cat << 'EOF' > normalize_script.R
   library(preprocessCore)
   
   # Load input data matrix
   # Assumes the first column contains row names (e.g., gene IDs) and subsequent columns are sample data.
   # Adjust 'sep', 'header', and 'row.names' based on your actual input file format.
   input_data <- read.table("input_data.tsv", sep="\t", header=TRUE, row.names=1)
   data_matrix <- as.matrix(input_data)
   
   # Load the defined reference distribution
   # Assumes it's a single-column file without a header.
   reference_dist <- as.vector(read.table("reference_distribution.tsv", header=FALSE)[,1])
   
   # Perform quantile normalization with respect to the defined reference distribution
   # The 'target.distribution' argument ensures normalization to a specific reference.
   normalized_matrix <- normalize.quantiles.use.target(data_matrix, target.distribution=reference_dist)
   
   # Restore column and row names to the normalized matrix
   colnames(normalized_matrix) <- colnames(data_matrix)
   rownames(normalized_matrix) <- rownames(data_matrix)
   
   # Save the normalized data to an output file
   # 'col.names=NA' is used to prevent writing a blank column name for the row names column.
   write.table(normalized_matrix, "normalized_counts.tsv", sep="\t", quote=FALSE, col.names=NA)
   EOF
   
   # Execute the R script
   Rscript normalize_script.R

   Copy

Tools Used