GSE125125 Processing Pipeline

Publication

eLife (2019) — PMID 30747709

Dataset

Overriding FUS autoregulation triggers gain-of-toxic dysfunctions in autophagy-lysosome axis and RNA metabolism

1. 1

   Sequenced reads filtered for low quality reads with skewer 0.2.2 (-Q 5)

   skewer v0.2.2 GitHub

   $ Bash example

   # Install skewer (if not already installed)
   # conda install -c bioconda skewer=0.2.2
   
   # Example command for quality filtering with skewer
   # Replace 'input_reads.fastq.gz' with your actual input file(s)
   # Replace 'output_prefix' with your desired output file prefix
   skewer -Q 5 -o output_prefix input_reads.fastq.gz

   Copy View on GitHub
2. 2

   RNA-seq read quantification was performed using Kallisto 0.44.0 (-b 50 –single -l 200 -s 20)

   kallisto v0.44.0 GitHub

   $ Bash example

   # Install kallisto (if not already installed)
   # conda install -c bioconda kallisto
   
   # Placeholder for kallisto index. Replace 'transcriptome.idx' with your actual index file
   # and 'transcripts.fasta' with your reference transcriptome FASTA file (e.g., from Ensembl, GENCODE).
   # kallisto index -i transcriptome.idx transcripts.fasta
   
   # Perform RNA-seq read quantification using Kallisto 0.44.0
   # Replace 'reads.fastq' with your input single-end FASTQ file.
   # Replace 'output_dir' with your desired output directory.
   kallisto quant -i transcriptome.idx \
                  -o output_dir \
                  --single \
                  -l 200 \
                  -s 20 \
                  -b 50 \
                  reads.fastq

   Copy View on GitHub
3. 3

   DEG calling was performed using Sleuth 0.28.1 (Control contrast)

   Sleuth v0.28.1

   $ Bash example

   # Install R if not already installed (example for Ubuntu)
   # sudo apt-get update && sudo apt-get install -y r-base
   
   # Install Bioconductor and Sleuth (if not already installed)
   # R -e 'if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")'
   # R -e 'BiocManager::install("sleuth")'
   # R -e 'install.packages("dplyr")' # Often used with sleuth
   # R -e 'install.packages("readr")' # Often used with sleuth
   
   # --- Placeholder for Kallisto output directories ---
   # Sleuth requires kallisto (or similar) quantification results.
   # These directories would typically contain 'abundance.h5' and 'run_info.json'.
   # For demonstration, we create empty directories.
   # In a real scenario, these would be generated by running Kallisto against a transcriptome index (e.g., GRCh38).
   mkdir -p kallisto_output/sample1_control
   mkdir -p kallisto_output/sample2_control
   mkdir -p kallisto_output/sample3_treated
   mkdir -p kallisto_output/sample4_treated
   
   # Create a dummy sample-to-condition file (s2c.tsv)
   # This file links sample IDs to their experimental conditions and kallisto output paths.
   # The 'control contrast' implies comparing a treatment group against a control group.
   echo -e "sample\tcondition\tpath" > s2c.tsv
   echo -e "sample1_control\tcontrol\tkallisto_output/sample1_control" >> s2c.tsv
   echo -e "sample2_control\tcontrol\tkallisto_output/sample2_control" >> s2c.tsv
   echo -e "sample3_treated\ttreated\tkallisto_output/sample3_treated" >> s2c.tsv
   echo -e "sample4_treated\ttreated\tkallisto_output/sample4_treated" >> s2c.tsv
   
   # Create an R script to perform DEG calling with Sleuth
   cat << 'EOF' > run_sleuth.R
   library(sleuth)
   library(dplyr)
   library(readr)
   
   # Load sample-to-condition information
   s2c <- read_tsv("s2c.tsv")
   
   # Ensure 'condition' is a factor and set 'control' as the reference level
   s2c$condition <- factor(s2c$condition, levels = c("control", "treated"))
   
   # Initialize sleuth object
   # 'extra_bootstrap_data = TRUE' and 'read_bootstrap_tpm = TRUE' are recommended for full Sleuth functionality
   so <- sleuth_prep(s2c, extra_bootstrap_data = TRUE, read_bootstrap_tpm = TRUE)
   
   # Define the full model (e.g., ~condition) to test for differences between conditions
   so <- sleuth_fit(so, ~condition, 'full')
   
   # Define the null model (e.g., ~1 for no effect)
   so <- sleuth_fit(so, ~1, 'null')
   
   # Perform Wald test for the 'conditiontreated' coefficient.
   # This directly tests the difference between 'treated' and 'control' (the 'control contrast').
   # The coefficient 'conditiontreated' represents the log2 fold change of 'treated' vs 'control'.
   so <- sleuth_wt(so, 'conditiontreated', 'full')
   
   # Get results for the specified contrast
   results_table <- sleuth_results(so, 'conditiontreated', test_type = 'wald', show_all = FALSE)
   
   # Filter results by q-value (FDR) and arrange by p-value
   results_table_filtered <- dplyr::filter(results_table, qval <= 0.05) %>%
     dplyr::arrange(pval)
   
   # Write all results and filtered results to TSV files
   write_tsv(results_table, "sleuth_deg_results_all.tsv")
   write_tsv(results_table_filtered, "sleuth_deg_results_q0.05.tsv")
   
   # Optional: Save the sleuth object for further interactive analysis
   # save(so, file = "sleuth_object.RData")
   EOF
   
   # Execute the R script
   Rscript run_sleuth.R

   Copy