GSE299099 Processing Pipeline

Publication

Cell chemical biology (2026) — PMID 41534524

Dataset

Transcriptome changes in MCF7 cells after treatment with NONO ligands and controls

1. 1

   Transcript abundance was quantified using Salmon [v1.3.0] with GENCODE v37 annotation.

   Salmon v1.3.0

   $ Bash example

   salmon index -t hg19.fa i salmon_index -k 31

   Copy
2. 2

   Gene level quantification was performed using tximeta [v1.8.4].

   tximeta v1.8.4 GitHub

   $ Bash example

   # Install R and Bioconductor if not already present
   # R -e "if (!requireNamespace('BiocManager', quietly = TRUE)) install.packages('BiocManager')"
   # R -e "BiocManager::install('tximeta')"
   # R -e "BiocManager::install('readr')" # For read_tsv
   
   # Create a dummy samples.tsv file for demonstration
   # This file would typically contain sample IDs, condition, and paths to quantification directories
   echo -e "sample\tcondition\tquant_dir" > samples.tsv
   echo -e "sample1\tcontrol\t./salmon_quant_dir1" >> samples.tsv
   echo -e "sample2\ttreated\t./salmon_quant_dir2" >> samples.tsv
   
   # Placeholder for quantification directories (e.g., from Salmon, Kallisto, RSEM)
   # In a real scenario, these would be generated by a prior quantification step.
   # mkdir -p salmon_quant_dir1 salmon_quant_dir2
   # touch salmon_quant_dir1/quant.sf # Placeholder for Salmon output
   # touch salmon_quant_dir2/quant.sf # Placeholder for Salmon output
   
   # Reference dataset: The reference transcriptome and annotation used for the initial quantification
   # (e.g., Salmon, Kallisto) is implicitly used by tximeta to build or retrieve a TxDb object.
   # Example: Human genome assembly (e.g., GRCh38/hg38) and GENCODE annotation (e.g., v45).
   
   # R script to perform gene-level quantification using tximeta
   Rscript -e '
     library(tximeta)
     library(readr)
   
     # Read sample metadata
     coldata <- read_tsv("samples.tsv")
   
     # Define paths to quantification files (e.g., Salmon quant.sf)
     # tximeta expects a "files" column pointing to the quantification output
     coldata$files <- file.path(coldata$quant_dir, "quant.sf")
   
     # Import quantification data with metadata
     # tximeta will automatically detect the reference transcriptome if it was indexed with a known FASTA/GTF
     # and will try to build/load a TxDb object from Bioconductor AnnotationHub.
     se <- tximeta(coldata)
   
     # Summarize to gene level
     # This step requires a TxDb object, which tximeta tries to automatically create/load
     # based on the reference used for quantification.
     gse <- summarizeToGene(se)
   
     # Save results (e.g., gene-level counts and TPMs)
     # For demonstration, we extract counts and TPMs
     counts_matrix <- assays(gse)$counts
     tpm_matrix <- assays(gse)$abundance
   
     write.csv(counts_matrix, "gene_counts.csv")
     write.csv(tpm_matrix, "gene_tpm.csv")
   
     # Optionally, save the summarizedExperiment object for further analysis
     saveRDS(gse, "gene_level_summarized_experiment.rds")
   '

   Copy View on GitHub
3. 3

   Differential gene expression was analyzed by DESeq2 [v1.30.1]

   DESeq2 v1.30.1 GitHub

   $ Bash example

   # Install DESeq2 (R package) via Bioconda
   # conda install -c bioconda bioconductor-deseq2=1.30.1
   
   # Create a placeholder R script for DESeq2 analysis
   cat << 'EOF' > deseq2_analysis.R
   #!/usr/bin/env Rscript
   
   # Load DESeq2 library
   library(DESeq2)
   
   # --- Placeholder for input files ---
   # Replace with actual paths to your count matrix and sample information
   # The count matrix should have genes/features as rows and samples as columns.
   # The sample information file should have samples as rows and metadata (e.g., 'condition') as columns.
   count_matrix_file <- "counts.csv"
   sample_info_file <- "sample_info.csv"
   output_results_file <- "deseq2_results.csv"
   
   # --- Load data ---
   # Assuming counts are raw counts (integers) and samples are columns, genes are rows
   # Adjust 'row.names' and 'sep' as needed for your file format
   # For example, if your count matrix is tab-separated and has gene IDs in the first column:
   # count_data <- read.delim(count_matrix_file, row.names = 1, sep = "\t")
   count_data <- read.csv(count_matrix_file, row.names = 1)
   
   # Load sample information
   # For example, if your sample info is tab-separated and has sample IDs in the first column:
   # sample_info <- read.delim(sample_info_file, row.names = 1, sep = "\t")
   sample_info <- read.csv(sample_info_file, row.names = 1)
   
   # Ensure sample names match between count data and sample info
   # And ensure they are in the same order
   sample_info <- sample_info[colnames(count_data), , drop = FALSE]
   
   # --- Create DESeqDataSet object ---
   # Design formula: ~ condition is a common example.
   # Replace 'condition' with the actual column name in your sample_info that defines your experimental groups.
   # Ensure 'condition' is a factor.
   sample_info$condition <- factor(sample_info$condition)
   dds <- DESeqDataSetFromMatrix(countData = round(count_data), # DESeq2 expects integer counts
                                 colData = sample_info,
                                 design = ~ condition)
   
   # --- Run DESeq2 analysis ---
   message("Running DESeq2 analysis...")
   dds <- DESeq(dds)
   message("DESeq2 analysis complete.")
   
   # --- Extract results ---
   # Replace 'condition_groupA_vs_groupB' with your actual contrast.
   # For example, if 'condition' has levels 'treated' and 'control', you might use:
   # res <- results(dds, contrast=c("condition", "treated", "control"))
   # If you just want the default comparison (last level vs first level of the factor):
   res <- results(dds)
   
   # Order results by adjusted p-value
   res_ordered <- res[order(res$padj),]
   
   # --- Save results ---
   write.csv(as.data.frame(res_ordered), file = output_results_file)
   
   message(paste("DESeq2 results saved to:", output_results_file))
   EOF
   
   # Execute the R script
   Rscript deseq2_analysis.R

   Copy View on GitHub

Tools Used