For around 4000 individuals both metabolomics data and RNAseq/DNAm is available. Here we describe how we constructed a SummarizedExperiment-object containing metabolomics data for the overlapping individuals.
molgenis.connect(usrpwd=RP4_DB_USRPWD, url=RP4_DB)
subjects <- molgenis.get.all("subjects", verbose=FALSE)
dim(subjects)
head(subjects)
biobanks <- molgenis.get.all("biobanks", verbose=FALSE)
dim(biobanks)
head(biobanks)
samples <- molgenis.get.all("samples", verbose=FALSE)
dim(samples)
head(samples)
measurements <- molgenis.get.all("measurements", verbose=FALSE)
dim(measurements)
head(measurements)In order to map the RP4 metabolite data to the RP3 bios data a little processing of the metabolite bios_id is required.
subjects <- subset(subjects, !is.na(bios_id))
table(subjects$biobank)
subjects$real_bios_id <- ""
ll <- subjects$biobank == "LIFELINES"
subjects$real_bios_id[ll] <- subjects$bios_id[ll] ##everything fine
rs <- subjects$biobank == "ERF_ERGO"
subjects$real_bios_id[rs] <- subjects$bios_id[rs] ##everything fine
lls <- subjects$biobank == "LLS_PARTOFFS" | subjects$biobank == "LLS_SIBS"
subjects$real_bios_id[lls] <- paste0("LLS-", subjects$bios_id[lls]) ##a little bit pasting
ids <- getView("getIds", url=RP3_MDB, usrpwd=RP3_MDB_USRPWD)
ntr <- subjects$biobank == "VUNTR"
id <- match(subjects$bios_id[ntr], ids$person_id[ids$biobank_id == "NTR"]) ##not bios id but person id
head(subjects$bios_id[ntr])
subjects$real_bios_id[ntr] <- ids$bios_id[ids$biobank_id == "NTR"][id]
dim(subjects)
head(subjects$real_bios_id)SummarizedExperiment-objectNow we can select the overlapping individuals and create a SummarizedExperiment-object.
library(GenomicFeatures)
library(SummarizedExperiment)
##drop duplicates
subjects <- subset(subjects, real_bios_id != "LLS-589")
colnames(subjects)[2] <- "subject_id" ##I find the naming a bit confusing
colnames(samples)[3] <- "sample_id"
id <- match(subjects$subject_id, samples$subject_id)
colData <- cbind(subjects, samples[id,])
id <- match(colData$sample_id, measurements$sample_id)
measurements <- measurements[id,]
m <- measurements[,-c(1:2)]
m <- apply(m, 1, as.numeric)
rownames(m) <- colnames(measurements)[-c(1:2)]
##add bios uuid for linking to other omics data
mid <- match(colData$real_bios_id, ids$bios_id)
colnames(m) <- ids$uuid[mid]
metabolomicData <- SummarizedExperiment(assays=list(measurements=m),
colData=DataFrame(colData))
save(metabolomicData, file=file.path(VM_BASE_DATA, "RP4", "metabolomics_RP3RP4_overlap.RData"))The overlapping metabolite data is stored with in the colData the real_bios_id which can be used to link the data bios RP3 RNAseq or DNAm data. The data set can be loaded using data(metabolomicData).