Implementation of a normalisation method which corrects for sample specific gene length bias.

[FilipeJesus]

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.

[FilipeJesus]

@vetronz @tonyyzy @ajle96 let me know if any of you would be willing to review my PR (when it occurs)

[tonyyzy]

All the science bit sounds like voodoo magic to me! More than happy to review the code 😄

[FilipeJesus]

Thank you Tony. I will keep you updated with the science part once I understand it fully too 😄.

Also, @tonyyzy if you could help me with adding this new method into the cwl_writter class, when i get to it, that would be very helpful.

[vetronz]

I am not promising to understand it! But I too will have a read. I have limited time currently as working in ED and the days are pretty long 😑 but ill get to it

…

On Wed, Nov 13, 2019 at 9:24 PM FilipeJesus ***@***.***> wrote: @vetronz <https://github.com/vetronz> @tonyyzy <https://github.com/tonyyzy> @ajle96 <https://github.com/ajle96> let me know if any of you would be willing to review my PR (when it occurs) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2?email_source=notifications&email_token=AEFRRHIPCBTHHV6V4Z5IXW3QTPPVVA5CNFSM4JMZROZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED5Z4AI#issuecomment-553360897>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEFRRHIKYFJB5BSRMQQZ2BDQTPPVVANCNFSM4JMZROZA> .

[FilipeJesus]

Thank you Patrick. That would be very helpful as my understanding of the front end is still quite poor.