GSE21037 Processing Pipeline

Publication

Cell (2010) — PMID 21074045

Dataset

L1 retrotransposition in neurons is mediated by MeCP2

1. 1

   Gene-level signal estimates were derived from the CEL files by RMA-sketch normalization as a method in the apt-probeset-summarize program (see Yeo et al.

   apt-probeset-summarize vNot specified

   $ Bash example

   # Install Affymetrix Power Tools (APT)
   # conda install -c bioconda affymetrix-power-tools
   
   # Define input CEL files (replace with actual file paths).
   # Example: CEL_FILES="sample1.CEL sample2.CEL sample3.CEL"
   # Or, if all CEL files are in a directory:
   CEL_FILES=$(find /path/to/your/cel_files_directory -name "*.CEL" | tr '\n' ' ')
   
   # Define output directory
   OUTPUT_DIR="gene_level_estimates"
   mkdir -p "${OUTPUT_DIR}"
   
   # Define annotation file (replace with the correct annotation file for your array type and genome build).
   # This file provides the mapping from probesets to genes (e.g., a CDF file or a probeset-to-gene mapping file).
   # The specific file depends on the Affymetrix array used (e.g., HuGene-1_0-st-v1, HG-U133_Plus_2).
   # Example for a human array (e.g., HuGene-1_0-st-v1):
   # ANNOTATION_FILE="/path/to/HuGene-1_0-st-v1.r2.dt1.hg19.na33.transcript.csv"
   # Example for an older array:
   # ANNOTATION_FILE="/path/to/HG-U133_Plus_2.cdf"
   ANNOTATION_FILE="path/to/your_array_annotation.csv" # Placeholder for the specific array annotation file
   
   # Run apt-probeset-summarize with RMA-sketch normalization
   apt-probeset-summarize \
       --method rma-sketch \
       --cel-files "${CEL_FILES}" \
       --annotation-file "${ANNOTATION_FILE}" \
       --output-file "${OUTPUT_DIR}/gene_level_signals" \
       --log-file "${OUTPUT_DIR}/apt_log.txt"

   Copy
2. 2

   2007; PMID 7967047).

   BLAST (Inferred with models/gemini-2.5-flash) vlegacy BLAST (circa 2007) GitHub

   $ Bash example

   # Install BLAST (e.g., using conda)
   # conda install -c bioconda blast
   
   # --- Placeholder for reference database and query sequences ---
   # Replace 'reference.fasta' with your actual reference sequence file (e.g., genome, transcriptome).
   # Replace 'my_blast_db' with your desired database name.
   # This command creates a BLAST database from your reference FASTA file.
   makeblastdb -in reference.fasta -dbtype nucl -out my_blast_db
   
   # Replace 'query.fasta' with your actual query sequence file.
   
   # Run blastn (Nucleotide BLAST)
   # -query: Input query file
   # -db: BLAST database name
   # -out: Output file in tabular format
   # -outfmt 6: Tabular output format (standard for parsing)
   # -num_threads: Number of CPU threads to use
   blastn -query query.fasta -db my_blast_db -out blastn_results.tsv -outfmt 6 -num_threads 8

   Copy View on GitHub
3. 3

   Hierarchical clustering of the full dataset by probeset values was performed by complete linkage using Euclidean distance as a similarity metric in Matlab.

   MATLAB vNot specified (Standard Library)

   $ Bash example

   #!/bin/bash
   
   # Input data file (replace with your actual path and format)
   # Example: probeset_data.csv (assuming comma-separated values)
   INPUT_DATA="probeset_data.csv"
   OUTPUT_CLUSTERING_MATRIX="hierarchical_clustering_results.csv"
   OUTPUT_DENDROGRAM_PNG="dendrogram.png"
   
   # Create a temporary MATLAB script for clustering
   cat << EOF > run_clustering.m
   % MATLAB script for hierarchical clustering
   % Load data (adjust 'readmatrix' based on your actual data format, e.g., readtable, then table2array)
   try
       data = readmatrix('$INPUT_DATA');
   catch ME
       disp(['Error loading data: ', ME.message]);
       exit(1);
   end
   
   % Calculate pairwise Euclidean distances
   Y = pdist(data, 'euclidean');
   
   % Perform hierarchical clustering with complete linkage
   Z = linkage(Y, 'complete');
   
   % Save the clustering result (linkage matrix)
   writematrix(Z, '$OUTPUT_CLUSTERING_MATRIX');
   
   % Optional: Generate and save a dendrogram image
   % This might require a display server if running headless, or MATLAB's 'batch' mode
   % might handle it if 'figure('visible','off')' is used. For robust headless operation,
   % ensure Xvfb is running or consider alternative plotting libraries.
   try
       figure('visible','off'); % Create a figure without displaying it
       dendrogram(Z);
       title('Hierarchical Clustering Dendrogram');
       saveas(gcf, '$OUTPUT_DENDROGRAM_PNG');
       close(gcf); % Close the figure
   catch ME
       disp(['Warning: Could not generate dendrogram image. ', ME.message]);
   end
   
   exit(0); % Exit MATLAB successfully
   EOF
   
   # Execute the MATLAB script in batch mode
   # Assumes MATLAB is installed and its executable is in the system PATH.
   matlab -batch "run_clustering" -logfile matlab_clustering.log
   
   # Clean up the temporary MATLAB script
   rm run_clustering.m

   Copy