Gene ontology term comparisons between species

Question

[ibasan]

Dear all, I did an orthology analysis for four species and annotated the orthogroups which are unique for one of the species or species groups. Now I want to compare the found GO terms to each other if there are any statistical differences between the species. Furthermore, I want to compare the results between two groups of species (two species are sexual and the other two are asexual). The data looks like that:

Go term     Spec1   Spec2   Spec3   Spec4   Spec1&2 Spec3&4 database    
GO:0000003  23  0   4   5   1003    508 26755 
GO:0000228  2   0   0      5   190 96  7349 
GO:0000278  14  42  9   0   371 92  10200 
Total       39  42  13  10  1564    696 136800

The numbers represent the counts of the found GO term for each species or species group, respectively. The last column is the number from the reference database. Which test should i use to detect the GO terms with a significant difference between the species sets?

I really appreciate any help you can provide.

Answer 1

[Jean-Karim Heriche]

You have a contingency table so do a Chi-squared test.

[ibasan]

Dear Jean-Karim Heriche, could you please be more detailed. I want to get the significant differences for each GO term for each comparison. A Chi-squared test will show if there are any differences but not for what GO term and species, isn't it?

[Jean-Karim Heriche]

You first do a global test to check if there's independence between GO terms and species then if this leads you to reject the null hypothesis, do post-hoc tests to find where the differences are. Don't forget to correct for multiple testing. If you only care about 1/2 vs 3/4 then just test this.