simplifyBeta() missing functionality for aitchison distance

[msmcfarlin]

Hello,

I would like to use simplifyBeta() to extract Aitchison distance estimates for PCoA plotting. When I try running the function it returns an error that states ! Can't subset columns that don't exist. Columns A014, A021, A025, A051, A053, etc. don't exist.

simplifyBeta works with my metadata when extracting euclidean and bray-curtis estimates so I don't think it's an error with the sample names. I noticed in the beta diversity divnet object I am using that there isn't an aitchison distance metric. I also saw in the testBetaDiversity() code that it uses the 'fitted_z' object to calculate aitchison distance. So I was thinking of calculating the aitchison distance using some of that code, adding it to the beta divnet object, then using simplifyBeta to extract the estimates.

Does that sounds like it would work? Or maybe I'm missing a step in properly calculating the aitchison distance?

Here is the code I used...

#aitchison distance related functions from testBetaDiversity()
      get_aitchison_distance <- function(comp_matrix){
                  lr_matrix <- log_ratio(comp_matrix)
                  return(as.matrix(dist(lr_matrix)))}

      log_ratio <- function(comp_matrix){
                  lr_matrix <- log(comp_matrix)
                  lr_matrix <- lr_matrix -matrix(apply(lr_matrix,1, mean),ncol = 1)%*%matrix(1, ncol = ncol(lr_matrix))
                  return(lr_matrix)}

#calculate aitchison distance
      aitch_matrix <- get_aitchison_distance(beta_divnet_object$fitted_z)

#add column and rownames that match euclidean object
      colnames(aitch_matrix) <- colnames(beta_divnet_object$euclidean)
      rownames(aitch_matrix) <- rownames(beta_divnet_object$euclidean)

#add to beta divnet object
      beta_divnet_object$aitch_matrix <- aitch_matrix

#Run simplifyBeta() to extract estimates
      estimatesID <- simplifyBeta(beta_divnet_object,
                                                    phyloseq_object,
                                                    "aitch_matrix", "Sample_ID")

Thank you for your assistance!

[ailurophilia]

Hi Mike,

I apologize for the delay in my response. It looks like this is actually an issue that we started to patch and the pull request just didn't get merged, so I'll nip that in the bud shortly.

Your code looks good to me – do you get reasonable results?

Thanks!

Best, David

[ailurophilia]

Just updating to say that simplifyBeta should now accept Aitchison distances.

[msmcfarlin]

Hi David,

Thanks for the update! The results with my code seemed reasonable though I'll check with the updated simplifyBeta() function.

Thanks, Mike