search

rivasiker / autocoalhmm

Automated pipeline for running CoalHMM
11 stars 2 forks source link

explanation of the incongruence #11

Open yozeng opened 3 weeks ago

yozeng commented 3 weeks ago

Hi @rivasiker , If the proportion of inconsistent topology estimated by autocoalhmm is 30%, is this all due to ILS? Or is it possible that both ILS and introgression are contributing?

rivasiker commented 2 weeks ago

CoalHMM should ideally only be used when introgression is not present. In case there is introgression, you might be able to pick it from the posterior decoding because the two states incongruent with the species tree topology might be asymmetric, i.e., one of them might be more prevalent than the other. This is not expected if there is only ILS.

yozeng commented 2 weeks ago

Thank you for your reply! I asked this question because I've seen the work on distinguishing between ILS and introgression in your articles. So, is it right that the incongruent signals caused by introgression can also be detected by CoalHMM?

rivasiker commented 2 weeks ago

Yes, but the effect of introgression in the results produced by CoalHMM has never been systematically tested, so they should be interpreted with care. If there is introgression in the real data, the CoalHMM model would be misspecified and could lead to unforeseen biases in both the parameter estimation and the posterior decoding.

yozeng commented 1 week ago

Thanks! I've got it. By the way, I want to know whether the following topological correspondences are correct, especially V2 and V3: V0/V1: (((sp1, sp2), sp3), sp4) V2: (((sp1, sp3), sp2), sp4) V3: (((sp2, sp3), sp1), sp4)

rivasiker commented 1 week ago

Correct!