Retrieve KEGG Brite pathway information

KEGG (Brite) pathways

Background

• Uniprot does not have good pathway annotation
• GO terms (obtained from Uniprot) are sometimes difficult to work with as they are too specific or unspecific
• GO terms have the (dis-) advantage of labelling a gene with many different functional classes
• what we often like to have instead is a simple overview about gene-pathway relationship
• simple means here that one gene is associated with only one or few descriptive functions
• KEGG (Brite) is a manually curated (not homology based) database that offers such information
• this example uses the Bioconductor package KEGGRest to retrieve pathways for the bacterium Bacillus subtilis.

Libraries and test data

• Install KEGGREST package from Bioconductor

BiocManager::install("KEGGREST")

• load required libraries

suppressPackageStartupMessages({
  library(KEGGREST)
  library(tidyverse)
})

Retrieve pathways

Starting with organism ID

• KEGG uses organism IDs, for Bacillus subtilis it is bsu
• using this ID, we can retrieve gene-pathway relationships using a premade R function
• internally it uses the keggLink function to find pathways and keggList to retrieve human readable pathway names
• it also trims some unnecessary text
• it can be used with organism ID (example: bsu) or gene ID (example: bsu:BSU00040)

source("../source/get_kegg_pathways.R")

• apply function to retrieve pathways by organism

df_kegg <- get_kegg_pathways(id = "bsu")
head(df_kegg)

Starting with gene ID

• we can supply one or more IDs but they need to have the organism tag in the front

df_kegg_genes <- get_kegg_pathways(id = c("bsu:BSU00040", "bsu:BSU20340"))
head(df_kegg_genes)

Results

• overview about most abundant pathways by locus_tag
• only top 10 pathways are shown

df_summary <- df_kegg %>%
  group_by(kegg_pathway) %>%
  summarize(count = n()) %>%
  arrange(desc(count)) %>%
  slice(1:10)

head(df_summary)

df_summary %>%
  slice(10:1) %>%
  ggplot(aes(x = count, y = fct_inorder(kegg_pathway))) +
  geom_col()

• how many genes are associated with multiple pathways?
• this anaylsis is the inverse of the previous one
• 524 genes are associated with only 1 pathway, 241 with 2, and so on

df_summary <- df_kegg %>%
  group_by(locus_tag) %>%
  summarize(genes_per_pathway = n()) %>%
  count(genes_per_pathway)

df_summary

df_summary %>%
  ggplot(aes(x = genes_per_pathway, y = n)) +
  geom_line() +
  geom_point()