I have a gene set containing a set of 1500 genes in which I did differential gene expression using limma voom pipeline. After this step I wanted to analyze a specific set of pathways the metabolism set from KEGG database.

hs_kegg_df <- msigdbr(species = "Homo sapiens") %>%
gs_cat == "C2", # This is to filter only to the C2 curated gene sets
gs_subcat == "CP:KEGG" # This is because we only want KEGG pathways

Then since I only want to analyze metabolism I did this (keep only the pathways envolved in metabolism):

hs_kegg_df_with_gs_source <- subset(hs_kegg_df, grepl("^hsa00|^hsa01|^hsa02", gs_exact_source))

My problem is that, for example, for the pathway KEGG arginine and proline metabolism the set contains 54 genes in which my kit only covers 4, and these for 4 genes are Differentially expressed yet when doing: ( I decided to do ORA because with 1500 genes GSEA is not recommended).

em <- enricher(genes, minGSSize=1, maxGSSize = 20,  universe = names(gene_list), TERM2GENE=hs_kegg_df,pAdjustMethod = "fdr")  

This set does not appear with significant p value. I also tried passing to the enricher just that pathway in particular, the p value is 1. Because in hypergeometric test, the background genes are the same as the as DGE genes in all pathways giving this way 1-0=1.

How can I analyze just one pathway in particular given this specific conditions?

These are the 4 genes DGE and included in my kit:

  gs_name                              gene_symbol
  <chr>                                <chr>      

And this are all the 59.

here all the genes dge and all the genes in pathway:

 genes_dge <- c("ALDH2", "ARG2", "NOS1", "ODC1")
 all_genes <- c("ACY1", "AGMAT", "ALDH18A1", "ALDH1B1", "ALDH2", "ALDH3A2", "ALDH4A1", "ALDH7A1", "ALDH9A1", "AMD1", "AOC1", "ARG1", "ARG2", "ASL", "ASS1", "AZIN2", "CKB", "CKM", "CKMT1A", "CKMT1B",s"CKMT2", "CPS1", "DAO", "GAMT", "GATM", "GLS", "GLS2", "GLUD1", "GLUD2", "GLUD2", "GLUL", "GOT1", "GOT2", "LAP3", "MAOA", "MAOB", "NAGS", "NOS1", "NOS2", "NOS3", "OAT", "ODC1", "OTC", "P4HA1", "P4HA2", "P4HA3", "PRODH", "PRODH2", "PYCR1", "PYCR2", "PYCR3", "PYCR3", "SAT1","SAT2", "SMS", "SRM") 
  • $\begingroup$ Thank you for the answer @M__ its samples from two different subtypes of human cancer. And its "C2". $\endgroup$ Aug 22, 2023 at 15:26

1 Answer 1


Two questions

  1. Significance
  2. Lack of consistency between enzymes (log2) within the arginine and proline metabolism pathway.

. 1. We've dealt with significance in a past post on Limma (it was not voom though). I'll try and find the post and link it here.

  1. The answer is straight forward: whilst all four enzymes include Arginine and proline metabolism pathway
  • only ARG and ODC1 share exactly the same pathways. Thus these should give very similar results.

  • NOS1 is a key enzyme in 6 different pathways.

  • ALDH2 is involved in 13 different pathways.

Therefore NOS1 and ALDH2 would have "confounders" (from an experimental perspective, not host metabolism) because they are more complex than singly arginine and proline metabolism pathways.

The other factor is in cancer patients the transcriptomics will get messed up, ALDH2 could go all over the place.

Prediction ARG = ORD < NOS1 < ALDH2

**ALDH2** aldehyde dehydrogenase 2 
Glycolysis / Gluconeogenesis
Ascorbate and aldarate metabolism
Fatty acid degradation
Valine, leucine and isoleucine degradation
Lysine degradation
**Arginine and proline metabolism**
Histidine metabolism
Tryptophan metabolism
beta-Alanine metabolism
Glycerolipid metabolism
Pyruvate metabolism
Pantothenate and CoA biosynthesis
**ARG** arginase
**Arginine and proline metabolism**
Glutathione metabolism
**NOS1** nitric oxide synthase 
Arginine biosynthesis
**Arginine and proline metabolism**
Calcium signaling pathway
Apelin signaling pathway
**ODC1** ornithine decarboxylase 1
**Arginine and proline metabolism**
Glutathione metabolism

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.