Phylogenetic tree file for BGM from command line

[DrewBarratt]

Hi hyphy Team,

I was wondering whether you could provide some assistance on the following -

I'm looking to incorporate the BGM tool into a co-evolution pipeline and was wondering how I could go about creating a suitable phylogenetic tree for this analysis using hyphy from the command line. Ideally, I would like this tree to be analogous to the one used in the BGM tool from the Datamonkey website for reproducibility. I have made a few attempts to write out the corresponding choices/arguments but had no success.

Any advice would be greatly appreciated.

Regards,

Drew

[spond]

Dear @DrewBarratt,

You really have two choices here

1). Build a "quick-and-dirty" neighbor joining tree in HyPhy. Incidentally, this option appears to be broken in the BGM analysis in the current release (--tree neighbor-joining option). I'll fix in the next release. 2). Build a tree in any package that can produce Newick trees and provide it to HyPhy (--tree path/to/file)

If the tree if included in the alignment file, it will be always used for the analysis.

Can you elaborate on what is/is not working for you at the moment and which version of HyPhy you are using?

Best, Sergei

[DrewBarratt]

Dear Sergei,

Thank you ever so much for getting back to me so promptly.

Yes, --tree neighbor-joining was my first thought - thanks in advance for addressing this in the future release. I guess what I'm asking is, what would be the correct command to generate a "quick and dirty" NJ tree independently from the BGM run as a temp workaround to --tree neighbor-joining? or is this just not possible?

when I run % hyphy nj -h to view options and their required values I am presented with this error:

Error: A recursive dependency error in _Variable::Compute; this is an HBL implementation bug; offending variable is 'MESSAGE_LOGGING'

and when I run nj --distanceChoice Distance formulae I am presented with

Error: 'formulae' is not a valid choice passed to 'Distance Computation' ChoiceList using redirected stdin input or keyword arguments. Valid choices are Distance formulae, Full likelihood, Use existing matrix in call to ChoiceList(distanceChoice, "Distance Computation", 1, SKIP_NONE, {{Distance formulae, Use one of the predefined distance measures based on data comparisons. Fast.}, {Full likelihood, Estimate distances using pairwise MLE. More choices but slow.}, {Use existing matrix, Load a HyPhy distance matrix for the data file}});

Is it just the case that the nj tool cannot be called directly from a command?

Although, I am able to generate an NJ tree when the values are entered manually through command interface, this doesn't really align itself with the pipeline im attempting to make.

I have already made an attempt to use other packages to create NJ and ML trees, however the subsequent BGM results are either inaccurate or BGM finds the input trees to be unbalanced.

e.g. BGM results using a hyphy generated NJ tree -

Site 1	Site 2	P [Site 1 <-> Site 2]	Subs (1,2,shared)
3	63	0.980	278, 157, 103
19	63	1.000	172, 157, 152
44	208	1.000	3, 2, 2
63	82	0.990	157, 215, 118
63	94	0.566	157, 2, 2	----

BGM results using another tree package -

Site 1	Site 2	P [Site 1 <-> Site 2]	Subs (1,2,shared)
44	208	0.985	2, 1, 1
63	217	0.986	2, 1, 1

visual inspection of the sequence confirms that the BGM results using the hyphy generated NJ tree are correct, hence wanting to continue using this appraoch.

These analyses were undertaken in 2.5.38

Thanks again,

[DrewBarratt]

Hi Sergei,

Just letting you know that I've managed to circumvent the above "Error" by using a heredoc approach within a bash script.

Thanks for the help and feel free to close this issue.