Create a small SummarizedExperiment from one dataset in curatedMetagenomicData (including ape phylogenetic tree)

smallSE.R

library(curatedMetagenomicData)
library(SummarizedExperiment)
library(tidyr)
library(ape)

simplifynodes <- TRUE

makeTaxTable <- function(fullnames){
  taxonomic.ranks <- c("Kingdom", "Phylum", "Class", "Order", 
                       "Family", "Genus", "Species", "Strain")
  fullnames %>% 
    gsub("[a-z]__", "", .) %>% 
    data_frame() %>% 
    separate(., ".", taxonomic.ranks, 
             sep = "\\|", fill = "right") %>% 
    as.matrix()
}

eset = curatedMetagenomicData("AsnicarF_2017.metaphlan_bugs_list.milk", 
                              dryrun = FALSE,
                              counts = TRUE)[[1]]

tax.table <- makeTaxTable(rownames(eset))

if(simplifynodes){
  rownames(eset) <- gsub(".+\\|", "", rownames(eset))
}

smallSE = SummarizedExperiment(assays = list(counts=exprs(eset)),
                               colData=pData(eset),
                               rowData=tax.table,
                               metadata=list(phylo=getMetaphlanTree(simplify=simplifynodes)))
  
summary(metadata(smallSE)$phylo$tip.label %in% rownames(smallSE))
summary(rownames(smallSE) %in% metadata(smallSE)$phylo$tip.label)