Figure out why "Canis lupus familiaris" 1:1 orthologs are reduced about orthogene HOT 3 CLOSED

Noticing that this is returning data where the Gene.Symbol is actually the human ortholog, not the dog gene name as we would expect:

 tar_genes <- all_genes(
            species = "Canis lupus familiaris",
            method = "babelgene",
            run_map_species = TRUE)

from orthogene.

I think this might be the root of it:

dog_orths=subset(babelgene:::orthologs_df,taxon_id==9615)
dog_orths

We can see that many genes (e.g. A1BG, A1CF) have the same gene symbol across species and are 1:1 in that sense. But there are multiple mappings from the dog symbol to the dog ensembl ID. This makes orthogene think theyre not 1:1 orthologs and thus drops them. But if all we care about is gene symbol to gene symbol, then we should keep them.

from orthogene.

This was exactly the issue! Fixed it by adding an extra step to drop_non121

 if(isTRUE(symbol_only)){
        gene_map <- data.table::as.data.table(gene_map)
        if("support_n" %in% names(gene_map)){
            data.table::setorderv(gene_map,"support_n",-1)
        }
        gene_map <- gene_map[,.SD[1],by=c("input_gene","ortholog_gene")] 
    } 

After rerunning report_orthologs, you get a near-perfect reconstruction of species similarity to humans. The exception being that dogs now appear more closely related to humans than mice (which is technically wrong, but close enough given that we're only looking at protein coding genes with 1:1 orths):

from orthogene.