gene family expansion statistical test

Question

[wd]

Hi

I want to run a statistical test to identify gene families (GFs) with a significant expansion/loss of genes between three species (A,B and C).

For each species I have the number of genes for each gene family:

           A      B     C      TOTAL
GF_1   50     50    50        150
GF_2  100    20    50        170
GF_3   7        2       1         10
GF_X ....

I searched in the available literature, and either Fisher, Chi-squared or a binomial test are proposed, but I can not find which one is the best to apply on my data?

Any help is much appreciated.

[Devon Ryan]

Why do you need to assign a p-value to this? I don't think that makes any sense in this context (unless there's heterogeneity within members of each species...but that would vary by gene family).

Answer 1

[wkk.215774]

We usually use fisher exact test to detect the expansion and contraction of gene families. But for phylogenomics analysis, l prefer using CAFE.

[Devon Ryan]

Again, what does a p-value even tell you in this context? Unless there's heterogeneity within the population a change in family size of any amount always constitutes a change. There's no additional coherent concept of significance that can be attached to that.