CPU and memory footprints of a few operations on RaggedExperiment objects from TCGA

TCGA_re.R

## ---------------------------------------------------------------------------------------------------------------------------------
library(curatedTCGAData)
library(TCGAutils)
library(RaggedExperiment)


## -----------------------------------------------------------------------------------------------------------------------------------------
cnvdry <-
  curatedTCGAData(assays = "CNVSNP",
                  version = "2.0.1",
                  dry.run = TRUE)
nrow(cnvdry)
sum(as.numeric(sub(" Mb", "", cnvdry$file_size)))


## -----------------------------------------------------------------------------------------------------------------
cnv <-
  curatedTCGAData(assays = "CNVSNP",
                  version = "2.0.1",
                  dry.run = FALSE)
cnvexplist <- experiments(cnv)
length(cnvexplist)
sum(sapply(cnvexplist, ncol))
print(object.size(cnvexplist), units = "auto")


## ----------------------------------------------------------------------------------------------------------------------------------------
mutsdry <-
  curatedTCGAData(assays = "Mutation",
                  version = "2.0.1",
                  dry.run = TRUE)
nrow(mutsdry)
sum(as.numeric(sub(" Mb", "", mutsdry$file_size)))


## ----------------------------------------------------------------------------------------------------------------
muts <-
  curatedTCGAData(assays = "Mutation",
                  version = "2.0.1",
                  dry.run = FALSE)
mutsexplist <- experiments(muts)
length(mutsexplist)
sum(sapply(mutsexplist, ncol))
print(object.size(mutsexplist), units = "auto")


## ------------------------------------------------------------------------------------------------------------------------------------
system.time(cnvsimp <- simplifyTCGA(cnv[, , "BRCA_CNVSNP-20160128"]))


## -----------------------------------------------------------------------------------------------------------------------------------------------
print(object.size(cnv[["BRCA_CNVSNP-20160128"]]), units="Mb")
print(object.size(cnvsimp[["BRCA_CNVSNP-20160128_simplified"]]), units="Mb")