## ---- include = FALSE---------------------------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE,
  warning = FALSE,
  message = FALSE,
  comment = "#>"
)

## ----install, eval=FALSE------------------------------------------------------
#  if (!requireNamespace("BiocManager", quietly = TRUE))
#      install.packages("BiocManager")
#  
#  BiocManager::install("msigdb")

## ----load-packages, message=FALSE---------------------------------------------
library(msigdb)
library(ExperimentHub)
library(GSEABase)

## ----get-msigdb---------------------------------------------------------------
eh = ExperimentHub()
query(eh , 'msigdb')

## ----download-msigdb-sym-id---------------------------------------------------
eh[['EH5421']]

## ----download-msigdb-sym-accessor---------------------------------------------
#metadata are displayed
msigdb.v7.2.hs.SYM(metadata = TRUE)
#data are loaded
msigdb.v7.2.hs.SYM()

## ----download-msigdb-sym-getMsigdb--------------------------------------------
#use the custom accessor to select a specific version of MSigDB
msigdb.v7.2.hs.SYM = getMsigdb('hs', 'SYM')
msigdb.v7.2.hs.SYM

## ----append-kegg--------------------------------------------------------------
msigdb.v7.2.hs.SYM = appendKEGG(msigdb.v7.2.hs.SYM)
msigdb.v7.2.hs.SYM

## ----process-gsc--------------------------------------------------------------
length(msigdb.v7.2.hs.SYM)

## ----access-gs----------------------------------------------------------------
gs = msigdb.v7.2.hs.SYM[[1000]]
gs
#get genes in the signature
geneIds(gs)
#get collection type
collectionType(gs)
#get MSigDB category
bcCategory(collectionType(gs))
#get MSigDB subcategory
bcSubCategory(collectionType(gs))
#get description
description(gs)
#get details
details(gs)

## ----summarise-gsc------------------------------------------------------------
#calculate the number of signatures in each category
table(sapply(lapply(msigdb.v7.2.hs.SYM, collectionType), bcCategory))
#calculate the number of signatures in each subcategory
table(sapply(lapply(msigdb.v7.2.hs.SYM, collectionType), bcSubCategory))
#plot the distribution of sizes
hist(sapply(lapply(msigdb.v7.2.hs.SYM, geneIds), length),
     main = 'MSigDB signature size distribution',
     xlab = 'Signature size')

## ----list-collections---------------------------------------------------------
listCollections(msigdb.v7.2.hs.SYM)
listSubCollections(msigdb.v7.2.hs.SYM)

## -----------------------------------------------------------------------------
#retrieeve the hallmarks gene sets
subsetCollection(msigdb.v7.2.hs.SYM, 'h')
#retrieve the biological processes category of gene ontology
subsetCollection(msigdb.v7.2.hs.SYM, 'c5', 'GO:BP')

## ----load-limma, message=FALSE------------------------------------------------
library(limma)

#create expression data
allg = unique(unlist(geneIds(msigdb.v7.2.hs.SYM)))
emat = matrix(0, nrow = length(allg), ncol = 6)
rownames(emat) = allg
colnames(emat) = paste0('sample', 1:6)
head(emat)

## ----subset-msigdb------------------------------------------------------------
#retrieve collections
hallmarks = subsetCollection(msigdb.v7.2.hs.SYM, 'h')
msigdb_ids = geneIds(hallmarks)

#convert gene sets into a list of gene indices
fry_indices = ids2indices(msigdb_ids, rownames(emat))
fry_indices[1:2]

## ----download-msig-sym-id-mouse-----------------------------------------------
msigdb.v7.2.mm.SYM = msigdb.v7.2.mm.SYM()
msigdb.v7.2.mm.SYM

## ----sessionInfo--------------------------------------------------------------
sessionInfo()