Produce a conversion table of human genes and mouse orthologs using bioMart

create_ortholog_table.R

## initially based on orthologsBioMART::findOrthologs from https://github.com/vitkl/orthologsBioMART
## boiled down to essentially reproduce and extend this table: ProjecTILs::Hs2Mm.convert.table

features <- c() # select human features to find orthologs for; e.g. rownames(Seurat::GetAssayData(SO, slot = "data", assay = "RNA"))

# use biomart to map orthologs
hs_mart <- biomaRt::useMart("ensembl", dataset="hsapiens_gene_ensembl")
#mm_mart <- biomaRt::useMart("ensembl", dataset="mmusculus_gene_ensembl") # not needed

hs_ids <- biomaRt::getBM(attributes = unique(c("hgnc_symbol", "ensembl_gene_id")),
                         filters = "hgnc_symbol",
                         values = features,
                         mart = hs_mart) %>%
  tibble::as_tibble() %>%
  dplyr::rename("Gene.HS" = 1, "Gene.stable.ID.HS" = 2) %>%
  dplyr::distinct() %>%
  dplyr::filter(trimws(Gene.HS) != "", trimws(Gene.stable.ID.HS) != "")

mm_orth <- biomaRt::getBM(attributes = c("ensembl_gene_id", "mmusculus_homolog_ensembl_gene", "mmusculus_homolog_associated_gene_name"),
                          filters = "ensembl_gene_id",
                          values = hs_ids[,"Gene.stable.ID.HS",drop=T],
                          mart = hs_mart) %>%
  tibble::as_tibble() %>%
  dplyr::rename("Gene.stable.ID.HS" = 1, "Gene.stable.ID.MM" = 2, "Gene.MM" = 3) %>%
  dplyr::distinct() %>%
  dplyr::filter(trimws(Gene.MM) != "", trimws(Gene.stable.ID.MM) != "")

table <-
  hs_ids %>%
  dplyr::left_join(mm_orth) %>%
  tidyr::drop_na() %>%
  dplyr::select(-Gene.stable.ID.MM) %>%
  dplyr::distinct(Gene.HS, Gene.MM, .keep_all = T) # keep an arbitrary Gene.stable.ID.HS for each match

# make a long data frame from alt.symbol of ProjecTILs::Hs2Mm.convert.table
pt_table <-
  ProjecTILs::Hs2Mm.convert.table %>%
  dplyr::mutate(Alt.symbol = stringr::str_split(Alt.symbol, ",")) %>%
  dplyr::mutate(Alt.symbol.HS = stringr::str_split(Alt.symbol.HS, ",")) %>%
  tidyr::unnest(Alt.symbol) %>%
  tidyr::unnest(Alt.symbol.HS) %>%
  dplyr::mutate(Gene.stable.ID.HS = ifelse(Alt.symbol.HS == Gene.HS, Gene.stable.ID.HS, NA)) %>%
  dplyr::select(-Gene.HS, -Gene.MM) %>%
  dplyr::rename("Gene.HS" = Alt.symbol.HS, "Gene.MM" = Alt.symbol) %>%
  tibble::as_tibble()

# bind rows and make distinct
ortholog_table <-
  hs_ids %>%
  dplyr::left_join(mm_orth) %>%
  dplyr::select(-Gene.stable.ID.MM) %>%
  dplyr::bind_rows(pt_table) %>%
  tidyr::drop_na() %>%
  dplyr::distinct(Gene.HS, Gene.MM, .keep_all = T) # keep an arbitrary Gene.stable.ID.HS for each match

multi_entries <- stack(table(ortholog_table$Gene.HS)) # FYI
# filter ortholog_table for features that actually exist in ref Seurat object (e.g. mouse) and in own object (e.g. human)
# then pass to ProjecTILs::make.projection()