Create a big SummarizedExperiment from curatedMetagenomicData (including phylogenetic tree)

bigSE.R

## A big SE

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()
}

esl = curatedMetagenomicData("*metaphlan*", counts=TRUE, dryrun = FALSE)
eset = mergeData(esl)

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

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

phy.tree <- getMetaphlanTree(simplify = simplifynodes)

bigSE <- SummarizedExperiment(assays = list(counts = exprs(eset)),
                              colData = pData(eset),
                              rowData = tax.table,
                              metadata=list(phylo=phy.tree))

summary(metadata(bigSE)$phylo$tip.label %in% rownames(bigSE))
summary(rownames(bigSE) %in% metadata(bigSE)$phylo$tip.label)