Retreiving KO data by assigning organism to bacteria using ko_to_kegg

[cmiko6492]

Is it possible to run species-specific KEGG pathway annotation on all bacteria, excluding human specific annotations?

Generic KO to KEGG pathway annotation (not specific to any organism)
daa_annotated_results_df <- pathway_annotation(pathway = "KO", daa_results_df = daa_results_df, ko_to_kegg = TRUE)

[cafferychen777]

Hi @cmiko6492, yes for the practical goal of excluding human/eukaryote-oriented pathway classes in bacterial/prokaryotic PICRUSt2 analysis, but not by setting organism = "bacteria" in pathway_annotation().

organism in pathway_annotation() maps to a single KEGG organism code, such as hsa for human or eco for E. coli. It cannot represent "all bacteria". So:

• organism = NULL gives the generic KO/reference pathway annotation.
• organism = "eco" gives an E. coli-specific pathway projection.
• There is currently no supported organism = "bacteria" option for all bacterial organisms.

For bacterial/prokaryotic data, the recommended workflow is to filter at the KO-to-KEGG pathway abundance conversion step. In current ggpicrust2, ko2kegg_abundance() has filter_for_prokaryotes = TRUE by default:

kegg_abundance <- ko2kegg_abundance(
  file = "pred_metagenome_unstrat.tsv",
  filter_for_prokaryotes = TRUE
)

daa_results_df <- pathway_daa(
  abundance = kegg_abundance,
  metadata = metadata,
  group = "your_group_column",
  daa_method = "LinDA"
)

daa_annotated_results_df <- pathway_annotation(
  pathway = "KEGG",
  daa_results_df = daa_results_df,
  ko_to_kegg = FALSE
)

That filter removes KEGG BRITE/non-pathway buckets and pathway classes that are not appropriate for prokaryotic analyses, including organismal systems and most human disease categories such as cancer, immune disease, nervous system, endocrine system, and endocrine/metabolic disease pathways. It intentionally keeps pathways that can still be relevant for microbial studies, such as bacterial infection and antimicrobial resistance pathways.

If you specifically need the richer KEGG API annotation columns such as pathway_class and pathway_map for significant pathways, you can still annotate after DAA with generic KEGG information:

daa_annotated_results_df <- pathway_annotation(
  pathway = "KO",
  daa_results_df = daa_results_df,
  ko_to_kegg = TRUE,
  organism = NULL
)

I would not use organism = "hsa" for bacterial samples. Also, organism = "eco" should only be used when you specifically want the E. coli pathway projection, not as a general bacterial annotation.

Short version: use ko2kegg_abundance(..., filter_for_prokaryotes = TRUE) to exclude human/eukaryote-oriented pathway classes; pathway_annotation(organism = ...) is only for single-organism KEGG projection and cannot represent all bacteria.