HEATMAP based on DEG data

Question

[hellokwmin]

Hello,

I have transcriptome data comprised of three treatments: Control, cold stress, and drought stress.

I performed DEG analysis by comparing cold stress vs. Control (called as cold) and drought stress vs. Control (called as drought)

Using DEG dataset, I subsequntly conducted GO enrichment analysis. One of GO terms for biological process was (lets say) was cell wall biogenesis in cold DEG, and there are 10 genes belonging to this term. So, I extracted these 10 genes to compare the level of gene expression with drouhgt DEG.

But, in drought DEG set, only 5 genes are duplicated; probably other 5 genes I can find in raw dataset. My question is:

If I want to make a heatmap to compare 10 genes between cold and drought, should I mark 5 genes (which is not detected in drought DEG) as "NA" (so in scale bar the value might be expressed as 0)? ---- in this case, I only am using DEG dataset, not raw dataset.

or,

should I extract these 10 genes from rawdata and then, chage these raw values to log2_transformed value for presenting them? ---in this case, I can get information of intrested genes from GO analysis, but to draw heatmap, I can use raw dataset.

[Basti]

I do not understand what you consider "DEG dataset", you should go with normalized counts and calculate z-scores to draw your heatmap

Answer 1

[rfran010]

It sounds like you did "cold stress vs. Control" and "drought stress vs. Control" as separate DEG datasets (assuming a DEG dataset includes normalized counts and subsequent differential expression values).

I believe you should re-do your DEG dataset with all conditions together. This shouldn't drop any genes that are detected in at least one condition. Then use z-scores of normalized counts as suggested by @Basti.