search

ZainulArifin1 / Jurkat_Transcriptomics_Analysis

Code to generate figures in the Jurkat Transcriptomics Paper
0 stars 0 forks source link

Normalization Issue #1

Open siegelna opened 1 month ago

siegelna commented 1 month ago

It appears the counts were normalized twice for this analysis.

https://github.com/ZainulArifin1/Jurkat_Transcriptomics_Analysis/blame/c5acc079ec9b6c3023b12637fc0f0c57ac1fd20d/Peter_Jurkat_Figures_SVG.rmd#L146-L159

dds <- estimateSizeFactors(dds) #Normalizatin procedure
normalized_counts <- counts(dds, normalized=TRUE)
normalized_counts <- log2(normalized_counts+1)

I used EdgeR instead of DEseq2 but a glance at the normalized count matrices shows stark differences. I imagine a similar result would be observed with DEseq2. image

ZainulArifin1 commented 1 month ago

Thank you for your thorough comment. You are correct that I apply a further log-transformation after DESeq normalization. However, this step is solely for obtaining the normalized count matrix for gene-level visualization. The actual differential testing directly uses data normalized within DESeq. I hope this clears it up.

siegelna commented 1 month ago

Yes, it does. Thank you for the clarification.