Closed MarineAGLAVE closed 5 years ago
Hi @MarineAGLAVE, great contribution, again! Thanks a lot :)
Thank you much Marine! D.
Le 13 août 2019 à 17:11, MarineAGLAVE notifications@github.com a écrit :
I found a problem with DESeq2 for k-mers and limma-voom for genes. We don't specify what is the control condition and the condition to test, so R takes them in alphabetical order and the condition with the name closest to A becomes the control (which in my case, reverses all the logFoldChange in the results). This problem had to be observed for the differential analysis of genes with DESeq2 because the contrast is specified to correct that. I propose a correction for the differential analysis of k-mers with DESeq2 similar to the contrast for DESeq2 genes. And for Limma-voom genes, I propose to use the specific contrast system of this method (with design and contrast matrix).
You can view, comment on, or merge this pull request online at:
https://github.com/Transipedia/dekupl-run/pull/64
Commit Summary
• Correction of contrast in DESeq2 and Limma-voom File Changes
• M bin/DESeq2_diff_method.R (2) • M bin/limma-voom_ref_transcripts.R (16) Patch Links:
• https://github.com/Transipedia/dekupl-run/pull/64.patch • https://github.com/Transipedia/dekupl-run/pull/64.diff — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
--
Daniel Gautheret
Institute for Integrative Biology of the Cell
UMR 9198 - Universite Paris-Sud
Bat 400 - 91405 Orsay cedex
Tel Direct: 33 (0)1 69 82 62 38
http://rna.igmors.u-psud.fr
I found a problem with DESeq2 for k-mers and limma-voom for genes. We don't specify what is the control condition and the condition to test, so R takes them in alphabetical order and the condition with the name closest to A becomes the control (which in my case, reverses all the logFoldChange in the results). This problem had to be observed for the differential analysis of genes with DESeq2 because the contrast is specified to correct that. I propose a correction for the differential analysis of k-mers with DESeq2 similar to the contrast for DESeq2 genes. And for Limma-voom genes, I propose to use the specific contrast system of this method (with design and contrast matrix).