Should I use Z-score standardization or Min-Max scaling in biological data ?

Question

[shangguandong1996]

Hi, everyone :)

I am always confused about the choise of scale method. For example, when I plot heatmap, I have to make sure the count data are at the same scale, but I do not sure which method I should use, Z-scale or Max-min scale.

So I am wondering whether someone can give some advice about it. I know in the field of machine learning, the choice of scaling method may have some rule of thumb, but I may not sure in biology data, is there some preference ?

Best wishes

Guandong Shang

[DareDevil]

for plotting you can consider z score where as if you want to look at differentially expressed genes you may take normalized data

[Jean-Karim Heriche]

As usual it depends on the data and the objective. Z-score normalization may be more useful for statistical applications while min-max normalization may be better for visualization. Note that min-max normalization can be problematic if there are outliers.