Phangorn with binary state data generates error with modelTest and warning with optim.pml

[fpbarthel]

Hi there!

Thanks for the great phangorn package. I'm having some trouble using this with binary data and wonder why this is.

I have two issues:

1. Using my data I get the following error when I try to use modelTest to determine the optimal model:

Error in as.vector(x, "character") : 
cannot coerce type 'closure' to vector of type 'character'

2. Using my data I get the following "warning message" when I use optim.pml to fit a JC model.

Warning message:
In QM[lower.tri(QM)] <- x$Q :
  number of items to replace is not a multiple of replacement length

Here is a toy example I generated that reproduces the first issue and generates an error (rather than a warning) when trying to fit a JC model. Unfortunately I was not able to reproduce the warning message above using toy data. Nevertheless, using my actual data it seems to output a model with an improved log-likelihood, so perhaps I can ignore the warning message?

library(ape)
library(phangorn)

## Generate 10 taxa with 1000 binary (0/1) characters each
m <- sapply(sprintf("taxon%s",1:10), function(taxon) round(runif(1000, 0, 1)))
rownames(m) <- sprintf("char%s",1:nrow(m))
d <- dist(t(m), method = "binary")
tre <- bionj(d)

## Convert to phangorn format
mphy <- as.phyDat(t(m), type = "USER", levels = c(0,1))

## This fails with the error above (issue 1)
modelTest(mphy)

## Works fine
fit   <- pml(tre, mphy)

## Generates error below, different from the warning in issue 2
fitJC <- optim.pml(fit, TRUE)

The error when trying to fit the JC model on the toy data is as follows:

Error in if (!any(betahat < 0)) { : missing value where TRUE/FALSE needed
In addition: Warning message:
In if (!is.binary(tree)) tree <- multi2di(tree) :
the condition has length > 1 and only the first element will be used