which test is best for me

[suratsiz]

Hi,

So I have 8 genes and 10 sample (sample size may increase to 40 later).I have a dependent variable and these 8 genes will be my independent variables. So each time I will have only one dependent variable and 8 independent variable.

gene expression data consist of fractional numbers (e.g. 0.1 1.2 3.4 it can be minus as well). my dependent variable consists of +/- or 1,2,3 (I will do the test for two dependent variables).

So I want to investigate whether I can predict my samples' dependent variable property by just looking to their gene expression levels.

For example: lets say for the dependent variable 1,2,3 May I say gene 1 is expressed only in dep var 1 but not in 3.

I hope I am clear enough.:confused:

[JohnM]

Since you have independent variables that are continuous and a dependent variable that is more-or-less categorical (it is an integer), you should try discriminant analysis - this will allow you to see how well gene expression levels predict the property.

Here's a link:
http://www2.chass.ncsu.edu/garson/pa765/discrim.htm