Hello, I need some help with deciding the best way to present data. I have UNT and treated samples between a control and knock-out sample. I currently have the data presented as normalised gene counts in a heatmap, as I think this better represents what’s going on in my samples. Whenever I do log2FC I feel like it hides a lot of data and is not telling the true story. Is it okay to present normalised gene counts in a heatmap? Thank you.
2 answers
It is absolutely fine to present normalised gene counts. The most common way that we present data is median-scaled gene counts (gene-wise centering by the median). In R it would be something like:
centred.matrix <- assay(rlog.dds.object) - rowMedians(assay(rlog.dds.object))
In the Deseq2 workflow section 4.3 (Sample distances) you can see (with coding) how to normalise, calculate sample distances, and plot the heatmap "to assess overall similarity between samples: Which samples are similar to each other, which are different? Does this fit to the expectation from the experiment’s design?"
Log in to answer this question.
-
Help with edgeR analysis
written by rb_99 •
-
Surrogate variable analysis for paired RNA seq experiment
written by nhaus •
-
Extracting Clusters from a Dendrogram (RNA Seq Data)
written by robert.flynn.21 •
-
rna-seq/DEseq2 analysis - is it a linear combination problem?
written by aabbcc1 •
-
Understanding enrichment signal from TF ChIP-seq using phantompeakqualtools and plotFingerprint(dee…
written by ahnje770 •
-
creating box plot from DESeq2 result
written by aswanikrishnap1994 •
-
How to process RNA-seq data for a heatmap in R
written by freakyfliss1 •
-
Normalised counts of control samples in gene expression matrix
written by Pranavathiyani G
-
Analyzing RNAseq data using pheatmap (R)
written by Randell •
-
what does rsem.genesnormalised_result files in tcga database suggest
written by bt11d001 •
its arguably even better than presenting log2FC. you can ballpark a log2FC from counts, reverse isnt true.