I have a question for fgseaMultiLevel, when using the geneSets from msigdb, Homo sapiens: C2 category, should I filter the gene sets with respect to the genes present in my dataset using

filtered_geneSets93 <- lapply(geneSets, function(geneSet) {intersect(geneSet, unique_gene_ids93) })

where unique_gene_ids93 is the list of all the ensemble gene ids present in my counts matrix. Because when I am running fgseaMultiLevel as follows, I am getting different result if I use filtered_geneSets93 or geneSets entirely,

fgsea_results_one93 <- fgseaMultilevel(pathways = geneSets, stats = gene_stat_one93, minSize = 15, maxSize = 500)

which one would be better?