1
$\begingroup$

I have plant genotypes that have been classified as Invasive, Wild or Domestic. I am trying to investigate what explanatory variables (such as "leaf size", "width of petals" and other continuous variables) affect the probability of being classified a sone of the three groups. I therefore have to perform a multinomial logistic regression. However, I would like to account for phylogenetic non-independence using a kinship matrix.

Are there R tools that can perform such regression?

I considered lmekin for example but I don't think it manages to take a response that can take 3 different categories. Is there some function I would be interested in (in the ape package for example)?

$\endgroup$
1
$\begingroup$

If you willing to do Bayesian model fitting, this is likely possible but will take some tinkering to make sure the models are what you think they are. Stan (via rstan or brms), can fit multinomial logistic regressions (A demo from Thinkinator).

If you can convert these models to pure stan code, then you can incorporate phylogenetic/kinship/arbitrary correlation structure into the model following this example on github (or this one). I have had better luck getting phylogenetic regression to work correctly in pure stan than via convenience functions.

$\endgroup$
  • $\begingroup$ Thanks, that was helpful! Here is also an explanation for similar analysis with MCMCglmm $\endgroup$ – Remi.b Dec 12 '16 at 15:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.