/diff_splice_paper

Code for differential splicing comparison paper (Soneson, Matthes, et al.)

Primary LanguageR

Code for differential splicing comparison

This repository contains the code for the following paper comparing various methods for differential isoform usage (differential splicing) detection:

  • C Soneson*, KL Matthes*, M Nowicka, CW Law & MD Robinson: Differential transcript usage from RNA-seq data: isoform pre-filtering improves performance of count-based methods. Genome Biology 17:12 (2016).

One goal of the paper is to contrast different types of differential splicing detection methods (assembly-based, event-based, counting bin-based). Moreover, focussing on the counting-bin based methods, we compare several different ways of defining the counting bins, while keeping the inference engine (DEXSeq) fixed. To evaluate the effect of transcriptome complexity, we perform simulation studies based on both human and fruit fly.

The structure of this repository is as follows:

  • drosophila

    • reference_files/: Directory containing the mean-dispersion estimates used for the simulation
    • figures/: Directory containing some figures illustrating the fruit fly simulation
    • run_analysis_drosophila5.sh: Bash script to invoke all analysis steps for the fruit fly simulation.

  • hsapiens

    • reference_files/: Directory containing the mean-dispersion estimates used for the simulation
    • figures/: directory containing some figures illustrating the human simulation
    • run_analysis_human5.sh: Bash script to invoke all analysis steps for the human simulation.

  • software

    • Rcode/: directory containing the R scripts that are called from any of the bash scripts to generate the results shown in the paper.

  • manuscript_figures:

    • scripts/: Directory containing scripts to generate the result visualizations shown in the paper
    • figures/: Directory containing figures generated by the scripts

To run the bash scripts, several preparation steps are necessary: