Get REACTOME terms, translate to mouse

getREACTOME.R

# Retrieve REACTOME terms for human directly from the website, then map to mouse.
# Pull human to mouse mappings via biomaRt.

library(biomaRt)
library(data.table)
library(magrittr)
library(rtracklayer)
library(tidyverse)

options(timeout=999)
ensembl_version <- 101
reactome_version <- 85

# GTF files as gene_id to gene_name lookup
genes_human <-
  rtracklayer::import(paste0("https://ftp.ensembl.org/pub/release-", ensembl_version, "/gtf/homo_sapiens/Homo_sapiens.GRCh38.101.chr.gtf.gz")) %>%
  data.frame %>% filter(type=="gene") %>%
  dplyr::select(gene_id, gene_name, gene_biotype)

genes_mouse <-
  rtracklayer::import(paste0("https://ftp.ensembl.org/pub/release-", ensembl_version, "/gtf/mus_musculus/Mus_musculus.GRCm38.101.chr.gtf.gz")) %>%
  data.frame %>% filter(type=="gene") %>%
  dplyr::select(gene_id, gene_name, gene_biotype)

mart_human <- biomaRt::useEnsembl("genes", dataset="hsapiens_gene_ensembl", version=ensembl_version)
mart_mouse <- biomaRt::useEnsembl("genes", dataset="mmusculus_gene_ensembl", version=ensembl_version)

human_mouse <-
  biomaRt::getLDS(attributes=c("ensembl_gene_id", "hgnc_symbol"), 
                  attributesL=c("ensembl_gene_id", "mgi_symbol"), 
                  mart=mart_human, martL=mart_mouse,
                  uniqueRows=TRUE) %>%
  magrittr::set_colnames(c("human_id", "human_name", "mouse_id", "mouse_name")) %>%
  unique

# Get REACTOME
url_reactome <- paste0("https://reactome.org/download/", reactome_version, "/ReactomePathways.gmt.zip")
temp <- tempfile()
download.file(url_reactome, temp)
reactome_human_list <- fgsea::gmtPathways(unz(temp, "ReactomePathways.gmt"))
unlink(temp)

# as data.frame
reactome_human <- 
  lapply(names(reactome_human_list), function(x){
    data.frame(pathway=x, gene_name=reactome_human_list[[x]], check.names=FALSE)
  }) %>% 
  do.call(rbind, .) %>%
  dplyr::left_join(x=., y=genes_human, by=c("gene_name"), relationship = "many-to-many") %>%
  dplyr::filter(!is.na(gene_name) & !gene_name=="" & !is.na(gene_id) & !gene_id=="") %>%
  dplyr::mutate(gene=paste(gene_id, gene_name, sep="_")) %>%
  dplyr::select(pathway, gene) %>%
  unique

# map human to mouse
reactome_mouse <- 
  reactome_human %>%
  dplyr::left_join(x=., 
                   y=human_mouse %>% dplyr::mutate(gene=paste0(human_id, "_", human_name)),
                   by=c("gene"), relationship = "many-to-many") %>%
  dplyr::filter(!is.na(mouse_id) & !is.na(mouse_name) & !mouse_id=="" & !mouse_name=="") %>%
  dplyr::mutate(gene=paste0(mouse_id, "_", mouse_name)) %>%
  dplyr::select(pathway, gene) %>%
  unique

save_that <- c("genes_human", "genes_mouse", "human_mouse", "reactome_human", "reactome_mouse")

lapply(save_that, function(x){
  
  data.table::fwrite(x=get(x), file=paste0(x, ".txt.gz"), col.names=TRUE, row.names=FALSE, 
                     sep="\t", quote=FALSE, compress="gzip")
  NULL
  
}) %>% invisible