confusion

[enluo211]

When I was using this software package, I encountered some confusion, can you help me out?

[silastittes]

Happy to help! Looks like an image failed to upload maybe? Can I get some more details about the issue?

[enluo211]

Hello, sir. Regarding your 2020 article titled "rdmc: An Open Source R Package Implementing Convergent Adaptation Models of Lee and Coop (2017)
" published in G3 (Bethesda), I have encountered some difficulties. Specifically, while I have calculated the selected sites for the populations under each independent model, I am unsure how to determine which sites best fit a given model. Should this be evaluated based on likelihood ratios or some other criteria?
Rplots.pdf

[silastittes]

My understanding is that composite likelihoods don't come with the same statistical properties as proper likelihoods, so things like likelihood ratios, AICs, etc. should not be used. Lee and Coop (2017) use a parametric bootstrap approach, but I did not implement it in the rdmc package as it requires doing simulations using separate software. For my own work, where we were testing models on a lot of sweep regions, we just reported which site from which model has the max composite likelihood and how much larger it was than the next best (see Figure 5A).

[enluo211]

I understand it as follows: assuming there are 10 selected loci, each locus is evaluated using every model. The composite likelihood of the four models is then compared for each locus, and the locus is assigned to the model with the highest score. Is my understanding correct?

[rossibarra]

Yes!

…

On Mon, Dec 2, 2024 at 7:11 PM enluo211 ***@***.***> wrote: 3041f1.jpg (view on web) <https://github.com/user-attachments/assets/8beebd94-0208-4ac0-807b-bfd03d4aa4cd> F5.large.jpg (view on web) <https://github.com/user-attachments/assets/699e1ceb-f23f-413c-9d9d-5c5d0c0e04a8> I understand it as follows: assuming there are 10 selected loci, each locus is evaluated using every model. The composite likelihood of the four models is then compared for each locus, and the locus is assigned to the model with the highest score. Is my understanding correct? — Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AATZNYIDIDRDDQNTPWWOLWT2DUOPDAVCNFSM6AAAAABS23DUI2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMJTGQ2TKNBSHA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

--

____________ Jeffrey Ross-Ibarra (he/him) rilab.ucdavis.edu Dept. of Evolution and Ecology University of California Storer Hall, One Shields Ave Davis, CA 95616 Tel: 530-752-4565 @jrossibarra

[enluo211]

Thank you,sir