remove samples from phyloseq object that have less than n taxa?

[nextgenusfs]

I feel like there should be an easy answer to this question, but I can't seem to find it. All I would like to do is to drop samples from a phyloseq object where there are less than n number of taxa. Anybody know how to do this? ntaxa() yields total number of OTUs. I know that samples can be dropped with the prune_samples() function requiring a list of TRUE/FALSE, but not sure how to generate that list.

[vmikk]

Hello Jon!
You can try to use estimate_richness to count the number of OTUs per sample.

library(phyloseq)
data("esophagus")

phyloseq_richness_filter <- function(physeq, mintaxa = 10){
  sp <- estimate_richness(physeq, measures = "Observed")
  samples_to_keep <- rownames(sp)[ which(sp$Observed >= mintaxa) ]
  
  
  if(length(samples_to_keep) == 0){
    stop("All samples will be removed.\n")  
  }
  
  if(length(samples_to_keep) == nsamples(physeq)){
    cat("All samples will be preserved\n")
    res <- physeq
  }
  
  if(length(samples_to_keep) < nsamples(physeq)){
    res <- prune_samples(samples = samples_to_keep, x = physeq)
  }
 
  return(res)
}


phyloseq_richness_filter(esophagus, mintaxa = 30)
phyloseq_richness_filter(esophagus, mintaxa = 10)
phyloseq_richness_filter(esophagus, mintaxa = 100)

But keep in mind that estimate_richness will work only with count data (not proportions).
I made the pull-request to fix this behaviour (#891), but it was not merged yet.

HTH,
Vladimir

[nextgenusfs]

Thanks @vmikk! Hadn't thought of the species richness function - as I kept looking for something to sum() or count(), etc. Much appreciated!

-Jon