Background

Amazing research by Mandelboum et al. has recently found that many RNASeq analyses are producing false positive results due to bad normalisation practices. Popular normalisation methods like TMM, quantile and upper-quartile normalization and RLE do not correct for sample specific gene length bias.

In this issue we hope to introduce a new DGE "Analysis" method which utilises either cqn or EDASeq normalisation (perhaps two new method) which should provide users with an option which does correct for sample specific gene length bias.

Key Changes

• New cwl tool script for the new DGE module.
• New R script which applies either cqn or EDASeq normalisation followed differential expression analysis
• Add the new tool to the front end script
• Add the new tool as a method in the cwl_writer class

Tests

• All existing tests should pass.
• New unit and intergration tests for the new scripts developed.