pblischak
commented
1 year ago Hi Pedro -- this is definitely an interesting situation. Would you mind sharing a little bit more about your data set so that I can hopefully help out a bit more?
- How many individuals are sampled from each of the taxa in the triple you are testing here?
- How many sites are in your data matrix?
- What is the value of gamma in the bootstrap results that are significant?
My initial guess is that maybe only one of the individuals has a signal for hybridization or gene flow, so that when it is only represented once you don't detect hybridization. However, when you do bootstrapping, if it is included two or three times, then you do detect hybridization. Just a guess but hopefully we can get this figured out
Greetings @pblischak and @lkubatko,
I have been using HyDe as a tool for the analysis of possible hybridization between closely related taxa. The program is working as intended and I have had minimal problems trying to run it thanks to the detailed documentation and tutorials (thank you a lot!) but I've run into a situation that you might be able to help clarify.
There's one particular triple in our data set that we believe might be an hybrid or at least where some geneflow might have happened. Unfortunately it does not come as significant (although close) in a first HyDe analsysis. But since other metrics seem to point to an hybrid nature, I decided to run a bootstrap_hyde for this triple of interest. When running a bootstrap with 1000 replicates, almost 1/3 of replicates produce significant results (even after adjusting the P-value). I've also ploted the results as in the hyde_manuscript and distribution of gammas suggest likely uniform introgression.
Does this mean that if I ran my main analysis 1000 times I could get this triple as a significant result in around 300 runs? And to interprete this biologically? Is there "moderate" support for this triple in nature, or would it be totally incorrect to assume this?
Best regards, Pedro