Unless I misunderstood, there's a second avuncular test of relevance here, that IS significant. If you select (Ol,Pu) as the focal row, then there's a signifcant avuncular test (Z = -6.21). (This is just the inverse focal/uncle pair, i.e. focal lineage 5 and test lineage 10, from the same file).
The example you choose is non-signficant because the the uncle baseline is huge (Ma–Gr = 1450), which is likely because it's involved in other events (per the results, Gr-Ma introgression). When looking at it this way, we anchor on Ma (i.e. we ask if Ma+(Ol,Pu), 412 attachments, is larger than Ma+Gr, 1450 attachments, which it obviously isn't). But if we look at it the other way and anchor on (Ol,Pu), the baseline is smaller (252), so we get a significant results (i.e. we're asking if 412 attachments is significantly larger than 252, which it is).
I think this is nuance that is important. But I am also not certain that I have this right! If this is right, then I think the correct approach here is to actually look at both directions, and treat a connection as significant if either direction is signficant.
|
For users applying this workflow to their own data, the key requirement is that the final gene tree file must be a CSV with the first column called `gt`. |
Unless I misunderstood, there's a second avuncular test of relevance here, that IS significant. If you select (Ol,Pu) as the focal row, then there's a signifcant avuncular test (Z = -6.21). (This is just the inverse focal/uncle pair, i.e. focal lineage 5 and test lineage 10, from the same file).
The example you choose is non-signficant because the the uncle baseline is huge (Ma–Gr = 1450), which is likely because it's involved in other events (per the results, Gr-Ma introgression). When looking at it this way, we anchor on Ma (i.e. we ask if Ma+(Ol,Pu), 412 attachments, is larger than Ma+Gr, 1450 attachments, which it obviously isn't). But if we look at it the other way and anchor on (Ol,Pu), the baseline is smaller (252), so we get a significant results (i.e. we're asking if 412 attachments is significantly larger than 252, which it is).
I think this is nuance that is important. But I am also not certain that I have this right! If this is right, then I think the correct approach here is to actually look at both directions, and treat a connection as significant if either direction is signficant.
DAFT/Tutorial/tutorial.md
Line 114 in 2cc454a