I have a set of genes, their direction (up (1) or down (-1) regulated, p-values and adjusted p-values. I want to run limma's goana or kegg ontology.

By looking at the code I tried to construct my de to be able to run goana. As far as I understood the main columns are:

de$p.value # p-values
de$coef  # <0 or >0 so that gonna identifies if the gene is up or down regulated.

However it gives me some strange results, so I guess I did something wrong.

I have genes in ensembl ids, so I converted them into entrez_gene:

  mart =  useMart('ENSEMBL_MART_ENSEMBL',dataset='mmusculus_gene_ensembl',
                  host="may2012.archive.ensembl.org") 

  ensembl_ids = c(up.reg.all$Geneid,down.reg.all$Geneid)
  entrez_gene = getBM(
    filters= "ensembl_gene_id",
    attributes= c("ensembl_gene_id", "entrezgene"),
    values= ensembl_ids,
    mart = mart)
  entrez_gene = entrez_gene[match(ensembl_ids,entrez_gene$ensembl_gene_id),]

So, my data.frame is

> head(df_genes)
     p.value coef
1 0.66741542    1
2 0.02399576    1
3 0.04609667    1
4 0.99554092    1
5 0.91228805    1
6 0.24023346    1

And I run goana as follows:

go = goana(df_genes, geneid=entrez_gene$entrezgene, species ="Mm")

However, it gives me strange results, so that it seems I constructed my df_genes wrongly.

> head(go)
                                           Term Ont     N p.value coef P.p.value P.coef
GO:0006508                          proteolysis  BP  1466       0    0         1      1
GO:0007275 multicellular organismal development  BP  4580       0    0         1      1
GO:0007565                     female pregnancy  BP   112       0    0         1      1
GO:0007566                  embryo implantation  BP    53       0    0         1      1
GO:0008150                   biological_process  BP 23313       0    0         1      1
GO:0008152                    metabolic process  BP 10070       0    0         1      1