0
$\begingroup$

I have two main groups of mice: Control diet and high fat/high sucrose (HF/HS). Within each group, I have three genotypes. So for example: Control group: TG, KO, WT HF/HS group: TG, KO, WT

I have cycle threshold data for their gene expressions (10 results for each group). My question is: if I am investigating how the change in diet affects each genotype, shall I use one-way ANOVA to do so? And then paired t-test to just compare one genotype between the two main groups?

$\endgroup$
  • 2
    $\begingroup$ I'd cut "SPSS" from your title. It cuts down on readership. Evidently you're not asking for code, which is good, but mentioning any software is thus quite unneeded. $\endgroup$ – Nick Cox Feb 4 '18 at 16:07
0
$\begingroup$

I'm assuming that when you say you have "10 results for each group", you mean you have 10 independent mice in each group, and one measurement/mouse. Also, the assumption is that you don't have a priori predictions about specific comparisons you want to make.

If so, then you need to run a factoral analysis of variance (if you are using the menu system in SPSS, go to Analyze > General Linear Model > Univariate...; NOTE: That is GENERAL Linear Model, not GENERALIZED Linear Model).

Again, if you don't have specific comparisons that you are predicting a priori, then you might want to chose the option for post hoc contrasts to compare the 6 groups (2 diets X 3 genotypes). You might want to find out what post hoc method is most common in your discipline, but Tukey's HSD (Called just Tukey, in SPSS I believe) is pretty popular. The Bonferroni is also a pretty widely accepted method. These will make all pair-wise comparisons between your 6 groups, controlling for the inflation in your Type I error rate because you are making multiple comparisons and because you are doing this post hoc (i.e., after you already know that at least some of the groups are different because you have the ANOVA results).

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you for your help! Yes basically my objective is to look at the effects of a hormone on gene expression. 3 gene types for the hormone (Knockout, transgenic and normal wild type) are fed a control diet. 3 gene types for the hormone (knockout, transgenic and normal wild type) are fed an obesity inducing diet (HF/HS). I was thinking I could compare the gene expression fold changes as follows: WT-control vs. WT-HFHS using an unpaired t-test...then the same for KO-control vs. KO-HFHS and TG-control vs TG-HFHS. I have seen ANOVA being used a lot but was unsure whether it would be suitable! $\endgroup$ – Carrie V Schönfeld Feb 4 '18 at 16:53
  • $\begingroup$ Also, sorry it I didn't make it clear. I have 30 in each diet group, 10 for each genotype (WT, TG or KO) :) $\endgroup$ – Carrie V Schönfeld Feb 4 '18 at 16:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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