Differential Exon Usage Analysis with DEXSeq

DEXSeq Package

Analysis

00_Settings.sh: Various settings for pipeline

• Dependencies
  • create a conda environment with:

    conda create --name DEU
    conda activate DEU
    conda install sra-tools samtools star r-base
    

• Parameters
  • accesions: a list of accessions to pull
  • RNA_SEQ_DIR: This is where a directory containing RNA-seq FASTQ files is located
  • OUT_DIR: This is where the output directory should be located
  • MAX_CPUS: Number of threads to use
  • MAX_MEM_GB: Maximum memory in GB to use
  • HUMAN_GENOME_REF: path to .fna FASTA reference file
  • HUMAN_GENOME_GTF: path to the GTF annotation file
  • HUMAN_RNA_REF: unused
  • sjdbOverhang: Set to the RNA read-length - 1 more info

01_Download_Accessions.sh: A script to download a given set of accessions with sra-tools

• Output
  • FASTQs from given accesssions

02_Alignment.sh: A script to align the resulting FASTQs with STAR

• Output
  • BAMs for each sample; indexed and sorted

DEU_1.R: A script to perform Differential Exon Usage analysis with DEXSeq

• Input
  • BAMs
  • GTF for reference
• Output
  • DEXSeqReport: Standard DEXSeq report of all genes and their Differential Exon Usage (DEU) more info
  • dxr.rds: a R dataset object containing the DEXSeqResults object generated during analysis

DEU_2.R: A script to analyze the resulting DEXSeqResults .rds file

• Input
  • DEXSeqResults: object generated during previous analysis
• Output
  • DEU.xlsx: Contains two sheets:
    • Combined Genes: For each gene, lists the following:
      • gene: gene name / alias
      • gene_desc: short description of the gene
      • total_exons: total number of exons in the gene
      • pvalue_sig_exons < 0.05: number of exons with p-values that are significant in that gene given the threshold
      • pvalue_prop_sig: a proportion; $\frac{total_exons}{pvalue_sig_exons < 0.05}$
      • pvalue_na: number of exons without a pvalue
      • comb_pvalue: the combined exon pvalue after using Fisher's method
      • comb_pvalue_df: degrees of freedom when combining each exon pvalue
      • padj_sig_exons < 0.05: number of exons with adjusted p-values that are significant in that gene given the threshold
      • padj_prop_sig: a proportion; $\frac{total_exons}{padj_sig_exons < 0.05}$
      • padj_na: number of exons without a padj
      • comb_padj: the combined exon padj after using Fisher's method
      • comb_padj_df: degrees of freedom when combining each exon padj
    • All Genes: For each exon in each gene, lists the followning:
      • groupID: gene name / alias
      • gene_desc: short description of the gene
      • featureID: feature ID
      • exonBaseMean: exon base mean
      • dispersion: dispersion
      • stat: statistic
      • pvalue: p-value
      • padj: adjusted p-value
      • day_0: condition #1
      • day_100: condition #2
      • log2fold_day_100_day_0: the $log_2$ fold change of the conditions; in this case $\frac{day_100}{day_0}$