This is a follow up to my question posted here Processing proteomics data

In relevance to my research, I've been looking for proteomics data (control vs. diabetic) and I found a dataset in the article "Diabetes causes marked inhibition of mitochondrial metabolism in pancreatic β-cells" (figures from here: link).

I would like to know how to generate the following heat map from the normalized abundances reported.

Fold change heatmap: enter image description here

Abundance: enter image description here

Figure description:

Heat maps of relative mRNA and protein levels of the indicated genes in islets isolated from control and 2-week diabetic βV59M mice. Each box corresponds to a different animal. Colour indicates log2 fold-change. Heat map: TCA cycle. Red, proteins upregulated in diabetes; blue, proteins downregulated in diabetes. Grey, no change or not detected. PDK1, pyruvate dehydrogenase kinase 1. Abundance: Abundance of the indicated proteins, quantified by mass spectrometry, in islets isolated from control (black, Ctrl, n = 4) and 2-week diabetic βV59M (white, Diab, n = 4) mice.

I am not able to understand how the log2 fold-change is computed from the abundance. For instance, if we consider PDK1 (first panel) the abundance measure of control is ~50000 and diabetic is ~200000. Heat map says log2 fold-change (in the figure description added above). What is the fold-change that is referred to? From what is shown, I think this is not diabetic/control fold change; we see two panels (i.e one for control (4 samples) and the other for diabetic (4 samples)) and not a single panel. Since the legend is from (-2 to 2) I think these values are not log2 (abundance) as well.

Could someone please clarify how the heat map has been generated from the abundance measures? Basically, I want to know how to generate the first heatmap figure from abundance plots.

  • $\begingroup$ Briefly ... that is absolutely correct - its not clear. The 2 to -2 is log10 (standard log) there is no other explanation. The appendix of the paper used log2 because it stated this (it is weird because natural log is would be the obvious choice in that scenario). This heatmap must be standard log. The authors MUST state the transformation and they do not appear to have done this. I will forward heatmap stuff - as comments. $\endgroup$
    – M__
    Commented Apr 14, 2022 at 15:32
  • $\begingroup$ I do have sympathy with the OPs project - this is an unnecessarily difficult project, which is outside their control. $\endgroup$
    – M__
    Commented Apr 14, 2022 at 15:34
  • $\begingroup$ I got a reply yesterday from the corresponding author on the diabetic vs. control labels of the samples. $\endgroup$
    – Natasha
    Commented Apr 14, 2022 at 15:35
  • 1
    $\begingroup$ Thats good, happy the project is moving forward $\endgroup$
    – M__
    Commented Apr 14, 2022 at 15:37
  • 1
    $\begingroup$ Hi @M__ Thanks for the comment. I managed to plot the heatmap in python. But I am having some confusion regarding the figure description and the heatmap shown in the article. I'll share a detailed edit today. $\endgroup$
    – Natasha
    Commented Apr 19, 2022 at 4:44

1 Answer 1


The -2 to 2 should be the normalized version of log2 data (maybe mean or median normalization). This is commonly used in heatmaps to show downregulated value vs. upregulated value.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.