Statistical methods to determine descrimimant variables between two groups

Question

[Bioinfo]

Hello,

I want to know the different methods that can be used to determine discriminant variables between two groups

For example if i have two groups : Control and treated, how can i determine the discriminant variables between these two groups,

Also in case of multiple groups, what are the methods that can be used to determine the descriminant variables between these groups

Thank you

[brunobsouzaa]

What kind of experiment are you working with? Is this RNAseq? Please, provide a complete description on what you are looking for.

[Bioinfo]

Thank you for your response, no it's metabolomics data comparing Control and treated sampels

Answer 1

[Jean-Karim Heriche]

I would suggest to look into feature selection methods. It's unlikely you'll have a small number of variables (i.e. features) that would unambiguously separate the different groups (although this does occasionally happen and when it does, we often have some previous knowledge or intuition about these variables) but some variables are usually more relevant than others and in combination have discriminatory power. The general idea is to train a classifier and look at which variables the model relies on.