How to distinguish orthologs from paralogs

[leegene1992]

Dear Michael,

I know TOGA is a great tool that can distinguish orthologs from paralogs. But how to tell the ortholog from paralog in the output file orthology_classification.tsv? For example, the one2many gene group, is it possible to tell which one is ortholog and which is paralog? I could not find this information.

Best regards,
Jing

[MichaelHiller]

orthology_classification.tsv is only listing orthologs, not paralogs.

Here is an example of human SMN1 and SMN2 (human spec dup) having a single mouse ortholog (many2one).

ENSG00000205571	ENST00000626847.SMN2	reg_12253	ENST00000626847.SMN2.442	many2one
ENSG00000205571	ENST00000628696.SMN2	reg_12253	ENST00000628696.SMN2.442	many2one
ENSG00000205571	ENST00000380742.SMN2	reg_12253	ENST00000380742.SMN2.442	many2one
ENSG00000205571	ENST00000638794.SMN2	reg_12253	ENST00000638794.SMN2.442	many2one
ENSG00000205571	ENST00000380743.SMN2	reg_12253	ENST00000380743.SMN2.442	many2one
ENSG00000172062	ENST00000506163.SMN1	reg_12253	ENST00000506163.SMN1.631	many2one
ENSG00000172062	ENST00000503079.SMN1	reg_12253	ENST00000503079.SMN1.631	many2one
ENSG00000172062	ENST00000625245.SMN1	reg_12253	ENST00000625245.SMN1.631	many2one
ENSG00000172062	ENST00000380707.SMN1	reg_12253	ENST00000380707.SMN1.631	many2one

All SMN1 and SMN2 transcripts map to the mouse gene 'reg_12253'.
[in TOGA2, we will name this mouse gene better].

If you have a many2many, then reference genes will have >1 orthologs in the query.

[leegene1992]

Thanks. I got the point.
In your case SMN2 were duplicated from SMN1, however it seems TOGA1 did not give such information directly. Is it possible to include such information (duplication events, specifically the inferred ancestral and descendant relationships) in TOGA2 with synteny evidence?
Best.

[MichaelHiller]

I think both are a pretty much a tandem duplication. That means both copies have the same synteny.
In this case figuring out what is the original and what is the copy is difficult.
In principle, the chain ID gives an indication (lower number, higher ranking chain; likely larger synteny context). But this may not always be true.

And pls have a look at the TOGA paper Figure 1h (STRC guinea pig). There the original is lost and one of 3 copies is kept intact.