Retesting cast of simulate.binary.dated.tree.fgy output to phylo object

[gtng92]

Commit f0d4cf0 simplified cast of simulated.binary.dated.tree.fgy output to phylo object.
Later as I process the tree, the following error happens:

> obs.tree <- compartmental.model(theta, nsim=1, tips=100, model='sir.nondynamic')[[1]]
> obs.tree <- parse.input.tree(obs.tree, config)
Error in x$edge : $ operator is invalid for atomic vectors

> class(obs.tree)
[1] "phylo"

[ArtPoon]

@ArtPoon and @gtng92 to pick this apart later today

[ArtPoon]

Could not reproduce this issue on my Mac using f0d4cf0
and running through the example script example-compartmental.R

[gtng92]

No error produced on @0ldM4j0r 's computer, error reproduced on mine and @MathiasRenaud 's computer. Checking to see if it's because of an update to R. (Currently on 3.2.3, Faisal is on 3.4)

[gtng92]

Updated to R version 3.4.1, reinstalled Kaphi and colgem, but error is still there.

[ArtPoon]

That is really strange. Is it stochastic? Are you sure the config file is the same?

[gtng92]

On all the computers, I ran the example-compartmental.R script from the specified commit. They all rely on the example-compartmental.yaml file to load the config.
The only thing I changed each time I ran it was t.end=50 instead of t.end=1300 to eliminate the negative probability error (which is a separate issue).

[0ldM4j0r]

I re-ran the test on the machines that it worked on last week but for some reason they are giving the same error now (aka not working). This is really weird as I am sure I did it along side Tammy last week and it worked. So none of the workstations down stairs or the mac pass that test.

[0ldM4j0r]

I also tried to run it on Langley through ssh and that didn't worked. This is weird as it worked when running it through ssh from Hypnotoad

[gtng92]

Oops. Just noticed an indexing addition to the phylo cast from the commit.
Originally it was

phy <- tree[[1]]                   
class(phy) <- 'phylo'
return(phy)

The first line accesses the first index of tree, which is not a phylogenetic tree, but the first attribute of the rcolgem phylo object. Initially must have thought it outputted 1 phylogenetic tree. Now looks like:

class(tree) <- 'phylo'
return(tree)

Don't really get how it worked for some computers before but not for others with this indexing addition but at least it works now!