[R-sig-ME] MCMCglmm and massing records in relation to ainv

[Jarrod Hadfield]

Hi Szymek,

I'm aware of this problem.  The current work around is to create a  
record in the data.frame for the internal node and set the response to  
NA and the covariates (including fixed_eff) to anything you like.  The  
next MCMCghlmm update (probably next week) will fix this - or I can  
send you a version now if you like?

Cheers,

Jarrod




Quoting Szymek Drobniak <szymonmarian.drobniak at uzh.ch> on Thu, 13 Feb  
2014 14:45:52 +0100:

> Hello,
>
> I know that MCMCglmm - when using pedigree/phylogeny data - generates dummy
> entries for all phantom nodes in the tree that were introduced in the
> process of creating the pedigree. I'm currently working on a series of
> phylo-corrected models and have a problem: although my data does provide
> data for all tips of the tree I use - and both the data and tree have the
> same sets of species (anyway, otherwise it would generate another error) -
> but still I'm getting the warning  message that 1 additional record was
> generated. This prevents me from fitting residual structures of the kind
> rcov=~idh(fixed_eff):units since then it claims that the R-structure does
> not define unique residual for each data point.
>
> Does anyone has ideas how could I add this missing record to my data (i.e.
> what this stands for?)
>
> Cheers
> szymek
>
> --
>
> *Dr Szymon Drobniak*
> Anthropological Institute and Museum
> Office Y42-K-66
> University of Zürich - Irchel
> Winterthurerstrasse 190
> CH-8057 Zürich, Switzerland
>
> 	[[alternative HTML version deleted]]
>
>



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.