In vjcitn/MultiAssayExperiment: Software for the integration of multi-omics experiments in Bioconductor

knitr::opts_chunk$set(echo = TRUE, eval = FALSE)

Fetch from curatedMetagenomicData

Download the data as a list:

library(curatedMetagenomicData)
esetlist <- list(taxa = ZellerG_2014.metaphlan_bugs_list.stool()[, 1:10],
                 pathways = ZellerG_2014.pathabundance_relab.stool())
## species and strain-level taxa only:
esetlist$taxa <- esetlist$taxa[grep("s__", rownames(esetlist$taxa)), ]
## eliminate taxa-specific pathway contributions (only total pathway abundances):
esetlist$pathways <- esetlist$pathways[grep("g__", 
                                  rownames(esetlist$pathways), invert=TRUE), ]

Then create the MultiAssayExperiment:

library(MultiAssayExperiment)
mae = MultiAssayExperiment(experiments=esetlist, 
                           colData=colData(esetlist[[2]]))
mae
rownames(mae)

CCA with omicade4

library(omicade4)
maesub = mae[, mae$disease %in% c("cancer", "large_adenoma", "small_adenoma"), ]
##Get rid of rows that are all zero:
for (i in 1:length(experiments(maesub))){
  experiments(maesub)[[i]] <- experiments(maesub)[[i]][apply(assay(maesub)[[i]], 1, function(x) sum(x) > 0), ]
}
mcoin = mcia(assay(maesub))
plot(mcoin, phenovec=maesub$disease, sample.lab=FALSE)

iClusterPlus

Error, "system is computationally singular"

library(iClusterPlus)
datasets = assay(maesub)
datasets = lapply(datasets, t)
iclus = iCluster(datasets=datasets, k=5, lambda=c(0.2, 0.2))
plotiCluster(fit=iclus, label=maesub$disease)

sparse CCA

library(PMA)
mae2 <- mergeReplicates(intersectColumns(mae))
mycca = PMA::CCA(x=t(assay(mae)[[1]]), z=t(assay(mae)[[2]]))
mycca

Others to try

library(PMA)
library(made4)
library(MCIA)
##library(Rtopper)  #gene set analysis

vjcitn/MultiAssayExperiment documentation built on May 3, 2019, 6:13 p.m.