I want to make a heatmap for my rna seq experiment but im not sure I really understand how to. So I have fpkm, raw read counts for each sample and log2fold-change values for the DEGs between 2 conditions. What I want to end up with is something like the example figure below. What I dont understand is that the fold change values are comparing 2 groups of samples so where do the fold change values for each individual sample come from???? For example in the figure the first gene Gene C is shown to be upregulated in the first sample Sample27. So what is it upregulated in relation to??
Fold changes are derived by dividing the mean expression values in each group being compared, and then logging the result. For example, if the mean expression in corneal dystrophy cornea epithelium is 8 but it is only 2 in healthy cornea, then the log2 fold change is:
8/2 = 4
log2(4) = 2
The log2 FC is 2.
The heatmap represents scaled FPKM values, not fold-changes. We filter the FPKM expression matrix for differentialy expressed genes prior to heatmap generation, as we are typically only interested in clustering those genes that have been shown to have a statistically significant differential expression between the conditions that we are comparing. We then check the dendrogram structure and the heatmap shading to see if the genes do indeed segregate our conditions.
The actual shading in the heatmap, then, is related to scaled FPKM values (usually Z scores). Your Gene C in Sample 27, having a higher score and red shading, means that it is highly expressed when compared to the mean expression of all genes in all samples being plotted. To interpret it literally, it looks like its expression is ~1 standard deviation from the mean expression value. In Sample 1, however, the same gene has expression that is lower than the mean.
