I have a gene universe and list of genes of interest. Both are annotated with GO labels. But some of the GOs a very specific, another ones, contrary, are very broad. Could this fact introduce some bias in my GO enrichment analysis (with topGO package) ? Whether will be GO enrichment analysis on GO slims (created from initial set of available GOs in my gene universe) more correct in this aspect? Are there another advantages of using GO slims instead of raw (total amount of GOs available in  gene universe of the organism of interest) GOs population ?  Which pros and cons has each from the mentioned above approaches ?