Hi Muhammad, 

It looks to me you have a dataset which increases consistently as time goes by? With each time set being about an order of magnitude more?

Are you drawing these heat maps in R? Then I would suggest the package heatmaply, as it produces (what I think) is really nice heatmaps. But there are many other packages. 

Regarding your problem. If you plot your data you get only visualization of one set nicely, and the rest look either over-represented or underrepresented? 

Can I suggest a transformation? Maybe square root or log? This will scale your data quite nicely, still allowing visualization of each time point, and the dataset overall?

Hope this helps.

Andries

Hi Andries,

Thank you for your message!

Indeed you got it right. We have pre and post exposure observations in which the population has been increased appr. 10-fold in post exposure period for different bacterial groups. Therefore, I wanted to do heatmap to show the increase which is not working in relative percentage heatmap.

You suggestion of transformation sounds good to me! Indeed I can try it...If not, I will write you again for your suggestions! 

Thank you!

Hi Muhammad,

I think a log fold change is what you need. Calculate log(A1/A2) where A1 is the raw count of A at time T1 and A2, the raw count of A at time T2. This way, you'll be able to visualize positive and negative changes in abundance.