Ensembl Homology: Human orthologs of mouse genes labled "many2many" but correspond to a single mouse ortholog

Question

[adam.faranda]

The mouse gene: ENSMUSG00000004500 has 2 human homologs: ENSG00000249471, and ENSG00000083812. Looking the opposite direction, the indicated mouse gene is the only mouse homolog of these two human genes. I was hoping someone from Ensembl could clarify why is this labeled "many2many" instead of "one2many"?

I retrieved a table of mouse homologs for human genes (Ensembl v104, data set Human Genes GRCh39) using the following biomart query:

Attribute set: "homologs radio button" Attributes:

1. Gene.stable.ID
2. Mouse.gene.stable.ID
3. Mouse.homology.type
4. Homology related stats [ %identities, GOC, Alignment Coverage, and Confidence]

As I was reviewing the table, I found several records where the Mouse.homology.type was labeled "ortholog_many2many" where the homologous mouse gene had multiple human homologs, but those human homologs are only associated with the single mouse gene. Shouldn't these genes be labeled "one2many" as is true for most other similar cases:

R Filtering Code:

> with(
+   hom_2_mus %>%
+     select(
+       Human_ID=Gene.stable.ID, 
+       Mouse_ID=Mouse.gene.stable.ID, 
+       Mouse.homology.type
+     ) %>%
+     group_by(Human_ID) %>%
+     filter(n() == 1) %>%             # Human genes with only one mouse homolog
+     group_by(Mouse_ID) %>%
+     filter(n() > 1),                        # Mouse genes with multiple human homologs
+   table(Mouse.homology.type)
+ )
Mouse.homology.type
ortholog_many2many  ortholog_one2many 
                44               1477 

Answer 1

[Michal @Ensembl]

Hi Adam,

This particular mouse gene is the only homologue of the two human gene stable IDs, but has a listing of "many2many" still because this mouse gene has undergone a duplication event following the speciation. Although the homologous relationship is only shown between the gene stable IDs - this mouse gene has more than one protein coding gene attached, and so effectively the two human genes show a "many2many" relationship on the gene tree - with the 2 mouse proteins. It could be noted that the human side homologues include more than just those two genes, and those two gene stable IDs actually encode multiple protein coding genes.