Query biomaRt and the KEGG API to produce a table connecting KEGG pathways with gene names for mouse and human

retrieveKEGG.R

library(biomaRt)
library(data.table)
library(dplyr)
library(magrittr)

kegg <- sapply(c("mouse", "human"), function(x){
  
  if(x=="mouse"){
    
    dataset <- "mmusculus_gene_ensembl"
    u <- "mmu"
    gene_name <- "mgi_symbol"
    
  }
    
  if(x=="human"){
    
    dataset <- "hsapiens_gene_ensembl"
    u <- "hsa"
    gene_name <- "hgnc_symbol"
    
  }
  
  # Query biomaRt for a table connecting Ensembl gene id, mgi gene name and entrez gene id
  mart <- biomaRt::useEnsembl("ensembl", dataset=dataset, version=101)
  
  genes <- 
    biomaRt::getBM(attributes=c("ensembl_gene_id", gene_name, "entrezgene_id"), mart=mart) %>%
    dplyr::mutate(entrezgene_id=as.character(entrezgene_id)) %>%
    magrittr::set_colnames(c("gene_id", "gene_name", "entrezgene_id"))
  
  # Query KEGG API for a table connecting pathways and entrez gene id
  kegg_pathway2entrez <- 
    data.table::fread(paste0("https://rest.kegg.jp/link/", u, "/pathway"), data.table=FALSE, header=FALSE) %>%
    magrittr::set_colnames(c("pathway_id", "entrezgene_id")) %>%
    dplyr::mutate(entrezgene_id=as.character(gsub(paste0(u, ":"), "", entrezgene_id)),
                  pathway_id=gsub("path:", "", pathway_id))
  
  # Query KEGG API for a table connecting pathway id and pathway human-readable name
  kegg_pathway2name <-
    data.table::fread(paste0("https://rest.kegg.jp/list/pathway/", u), data.table=FALSE, header=FALSE) %>%
    magrittr::set_colnames(c("pathway_id", "pathway_name")) %>%
    dplyr::mutate(pathway_name=gsub(" - .*", "", pathway_name))
  
  # join everything
  kegg <- 
    dplyr::full_join(x=kegg_pathway2name, y=kegg_pathway2entrez, by="pathway_id") %>%
    dplyr::left_join(x=., y=genes, by="entrezgene_id") %>%
    dplyr::filter(!is.na(gene_id) & !is.na(gene_name))
  
  kegg

  }, simplify=FALSE)

head(kegg$mouse)
head(kegg$human)