I think you can use a simple hypergeometric test in R here, using function phyper. The random process is described using a drawing process of colored balls from a total. For example you could test if a specific phenotype or any phenotype is enriched in genes assigned to modules by co-expression network analysis. Chi-square test can be used to compare the complete distribution of phenotypes. Depending on your experiment design (e.g. differential expression), GSEA might also be applicable. Overall, standard statistics apply to this case. You always need to define suitable sub-populations of the genome so an enrichment analysis makes sense only in comparison of a sub-group vs. a global distribution.