boot parameter is silently ignored for Poisson glm, and summary yields an error

[Ax3man]

From the help, but with Poisson:

set.seed(123456)
tre = rtree(50)
x = rTrait(n=1,phy=tre)
X = cbind(rep(1,50),x)
y = rbinTrait(n=1,phy=tre, beta=c(-1,0.5), alpha=1 ,X=X)
dat = data.frame(trait01 = y, predictor = x)

fit = phyloglm(trait01~predictor,phy=tre,data=dat,boot=100, method = 'poisson_GEE')
summary(fit)

phyloglm doesn't do anything with the bootstrapping for Poisson models, but does run the model normally. Asking for the summary then returns an error from print.summary.phyloglm:

Error in if (x$bootNrep > 0) cat("      Parametric bootstrap results based on",  : 
  argument is of length zero

presumably since it is looking for bootstrap results that aren't actually there.

[lamho86]

You are right. We haven't implemented bootstrap for Poisson models yet. I will do it soon.

[Ax3man]

Not a problem of course! Just mentioning it here since it is perhaps not so clear to users what is going on when this fails.

[lauterbur]

Is the bootstrap for Poisson models still in the plan? It would be a great help for evaluating these models!

[ra-barber]

Was bootstrapping ever done for these models? I'm using the bootstrap for normal phylolm and it'd be great to plot model results alongside with same method. Or is it easy to code up from inspecting the function?

[lamho86]

Coding bootstrap for Poisson glm is straightforward. All you need is a function to simulate from the Poisson model (similar to the rTrait/rbinTrait functions). Then, you can integrate it into phyloglm just like the boostraps for logistic regression. I've got a lot on my plate, so I couldn't do it right now. Please do a pull request if you code up something and would like to add them to our package.