I am working with a non model organism and I performed a gene set enrichment test using topGO and GOstats. The analysis in both cases worked properly, but I would need an enrichment for a higher level of GO (less specific). Is someone aware of a way to select the annotation level to use for performing the enrichment with these two tools?

Thank you for your help.

What you would want to do is set conditional = FALSE (in GOstats), and whatever you do to achieve that result using topGO. That argument, when set to TRUE, will remove genes from more general GO terms when the gene gave rise to a significant result in a more specific term. Once you set it to FALSE, then each GO term is tested using all genes that are annotated to that term.

Since the hypergeometric test (once conditional is set to FALSE) isn't conditional on anything else, running GOstats or topGO on a subset of GO terms will not give results any different than running on all the terms, except for excluding some of the terms from consideration. In other words, subsetting to a set of less specific GO terms won't change any of your p-values, nor the ordering. It will just remove the more specific results from your list of terms, which you can easily do yourself by simply ignoring those terms in your output.