[R-sig-genetics] pegas::amova()

[Zhian Kamvar]

Hi Eric,

Can you provide a minimum working example and post your session information?

The following example works for me (adapted from https://github.com/emmanuelparadis/pegas/pull/2#issue-67437033):

library('pegas')
data(microbov)
set.seed(20150409)
mic20 <- microbov[sample(nInd(microbov), 20)]
microdf <- data.frame(other(mic20))
d <- dist(tab(mic20))
amova(d ~ spe/breed, data = microdf, nperm = 0)

SPE   <<- as.factor(microdf$spe)
BREED <<- as.factor(microdf$breed)
amova(d ~ SPE/BREED, nperm = 0)

> sessionInfo()
R version 3.2.1 (2015-06-18)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: OS X 10.10.4 (Yosemite)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] pegas_0.8-1    adegenet_2.0.0 ade4_1.7-2     ape_3.3       

loaded via a namespace (and not attached):
[1] Rcpp_0.11.6      spdep_0.5-88     plyr_1.8.3       LearnBayes_2.15  tools_3.2.1     
[6] boot_1.3-16      digest_0.6.8     nlme_3.1-120     gtable_0.1.2     lattice_0.20-31 
[11] Matrix_1.2-1     igraph_1.0.1     shiny_0.12.1     DBI_0.3.1        parallel_3.2.1  
[16] proto_0.3-10     coda_0.17-1      dplyr_0.4.2      stringr_1.0.0    grid_3.2.1      
[21] R6_2.1.0         sp_1.1-1         ggplot2_1.0.1    reshape2_1.4.1   seqinr_3.1-3    
[26] deldir_0.1-9     magrittr_1.5     scales_0.2.5     htmltools_0.2.6  MASS_7.3-41     
[31] splines_3.2.1    assertthat_0.1   mime_0.3         colorspace_1.2-6 xtable_1.7-4    
[36] httpuv_1.3.2     stringi_0.5-5    munsell_0.4.2  

The data frame implementation works now, so you can use your strata directly instead of pulling the levels from it:

my_strata <- strata(my_data)
amova_out <- pegas::amova(dists~l2/l1, data = my_strata, nperm=nperm, is.squared=T)

Additionally, poppr has a wrapper to ADE4's implementation of AMOVA that explicitly handles strata called poppr.amova.

Cheers,
Zhian 


> On Jul 20, 2015, at 12:16 , Eric Crandall <eric.d.crandall at gmail.com> wrote:
> 
> Hi Emmanuel or others,
> 
> 	I was directed here to ask a question about the amova function in pegas. Hopefully this is the right place.
> 
> 	 I have been using pegas::amova() in a function that loops over a database of species to calculate amova at up to three hierarchical levels. My function recently broke, apparently due to a recent update <https://cran.r-project.org/web/packages/pegas/NEWS> to Pegas - specifically this change:
> 
>   o amova() failed when the factors in the rhs of the formula were
>     in a data frame. Thanks to Zhian Kamvar for the fix.
> 
> As a workaround to for pegas:amova not being able to use a dataframe of factors, I was assigning the factors to the global environment:
> 
>     level1factor<<-as.factor(sp[[l1]])
>     level2factor<<-as.factor(sp[[l2]])
>     amova_out<-pegas::amova(dists~level2factor/level1factor, nperm=nperm, is.squared=T)
> 
> 
> Now, after the update, my function is throwing an error. The traceback goes
> 
> 4 stop("'bin' must be numeric or a factor") 
> 3 FUN(X[[i]], ...) 
> 2 lapply(gr, tabulate) 
> 1 pegas::amova(dists ~ level2factor/level1factor, nperm = nperm, 
>   is.squared = T) 
> 
> As far as I can tell, the objects that I am submitting to tabulate() in gr via lapply() ARE factors. 
> 
> When I try to reproduce the problem by running step by step through the pegas::amova function, everything works fine. 
> 
> So I am stumped for now. Does anyone have any ideas?
> 
> 
> Many thanks,
> 
> Eric Crandall
> 
> 
> 
> PS - I’m very happy to see strata being added to genind objects in adegenet. Will these eventually be usable by pegas::amova or a similar amova function?
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-genetics mailing list
> R-sig-genetics at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-genetics