merge pangenome graphs for different species

[joglekarp]

Hi

I am interested in merging individual pan-genomes generated for three species (with distinct ANI) within a genus. Is there a recommendation about whether it is okay to split paralogs in order to run panaroo-merge successfully?

Thanks

[gtonkinhill]

Hi,

It should be okay to split paralogs. One thing to note when merging graphs from different species is that Panaroo's ability to resolve very diverse gene clusters relies on gene synteny. Thus, if the gene synteny differs substantially between your species, some of the more diverse gene families may not be resolved.

[joglekarp]

Hi,

Thank you. This is very helpful. Just to be sure that as a novice I understand this process, merge graph is based on gene synteny and not sequence identity? Also can the --search_radius parameter be used during merging to resolve some of the "more diverse gene families" that you warn about? thx

[gtonkinhill]

Hi,

Not quite. It uses both synteny and sequence identity but very diverse genes that appear in very different locations in the two species may not be clustered together.

Adjusting the --search_radius might help a little bit but making it too large increases the chance that genes may be incorrectly clustered together. A better option might be to relax the initial clustering threshold using --threshold option.

I would suggest running the merge using the default settings initially and then investigating the resulting graph in Cytoscape to see if it looks reasonable. You could then test how sensitive the results were to changes in the parameter settings.

[joglekarp]

Thank you