Gene Loss and Forward Genomics

[shjenkins94]

Hello, I wanted to ask about using TOGA to identify gene loss.

I noticed that a 2018 paper from your lab ("A genomics approach reveals insights into the importance of gene losses for mammalian adaptations") used % intact for a "variant on the Forward Genomics approach," while a later 2022 paper ("Gene losses in the common vampire bat illuminate molecular adaptations to blood feeding") just looked at which genes were Lost/Uncertain Loss in the Vampire bat haplotypes but in other bats.

Are both methods still useful depending on the experiment design, or is just looking at Lost/Uncertain Loss genes enough to determine which genes are lost?

[MichaelHiller]

The 2018 paper was performing a screen for convergent gene losses that are associated with a trait. Here we thought we have more power if we use the %intact as a more quantitative measure for each gene-species pair. (From a technical point, this was pre-TOGA and we did not have the UL category).

The vampire work was looking at losses in a single lineage (vampires) and since we also ran RELAX and manually looked at all genes, we pulled out all those that are L or UL.

For a new screen, one could semi-quantitatively encode each gene as 0 = intact (or PI), 1 = UL and 2 = L.

[shjenkins94]

So for a study that's more similar to the 2018 paper (looking at convergent gene losses associated with a trait) would it be good to use the semi-quantitative method to identify lost genes and then look at those genes with RELAX?

[MichaelHiller]

yep, but in general I recommend to trying different ways. Robust results should be found with multiple 'versions'

[shjenkins94]

Also, what would M, PM, PG and N be encoded as? NA?