The ExpressionSet Data Object

Stefano Monti

Simple example of use of the R object ExpressionSet, ideal for the storage of gene expression or similarly structured omic data.

library(BS831)
library(Biobase) # The package with the ExpressionSet definition, among others

The ExpressionSet object

An expression set is a data object consisting of three entities: the expression matrix (exprs), the phenotye data (pData), and the feature data (fData).

Figure 1: ExpressionSet object

We upload an expression set already available. It corresponds to a subset of samples from a gene expression dataset of head and neck squamous carcinoma (HNSC) primary tissue samples from the TCGA project.

# Load toy dataset
data(HNSC_RNASeq_toy_ES)

## if BS831 not installed you can read the data from the Google Drive folder:
## HNSC_RNASeq_toy_ES <- readRDS(file.path(Sys.getenv("OMPATH"),"data/HNSC_RNASeq_toy_ES.rds"))

## rename for easier handling
hnsc <- HNSC_RNASeq_toy_ES

dim(hnsc)        # the expression data

## Features  Samples 
##    19844       25

dim(Biobase::pData(hnsc)) # the phenotypic annotation

## [1] 25  6

head(Biobase::pData(hnsc))

##                 bcr_patient_barcode bcr_sample_barcode tissue_type my_grade
## TCGA-CV-6962-11        TCGA-CV-6962    TCGA-CV-6962-11          AN       AN
## TCGA-CV-7432-11        TCGA-CV-7432    TCGA-CV-7432-11          AN       AN
## TCGA-CV-7177-11        TCGA-CV-7177    TCGA-CV-7177-11          AN       AN
## TCGA-CV-7438-11        TCGA-CV-7438    TCGA-CV-7438-11          AN       AN
## TCGA-CV-7440-11        TCGA-CV-7440    TCGA-CV-7440-11          AN       AN
## TCGA-BA-7269-01        TCGA-BA-7269    TCGA-BA-7269-01       Tumor       g1
##                  my_stage patient.anatomic_neoplasm_subdivision
## TCGA-CV-6962-11        AN                                larynx
## TCGA-CV-7432-11        AN                           oral cavity
## TCGA-CV-7177-11        AN                                larynx
## TCGA-CV-7438-11        AN                           oral tongue
## TCGA-CV-7440-11        AN                                larynx
## TCGA-BA-7269-01 stage iii                           oral tongue

dim(Biobase::fData(hnsc)) # the genes' annotation

## [1] 19844     3

head(Biobase::fData(hnsc))

##             entrez_id gene_symbol
## ?|100130426 100130426        <NA>
## ?|100133144 100133144        <NA>
## ?|100134869 100134869    UBE2Q2P2
## ?|10357         10357     HMGB1P1
## ?|10431         10431        <NA>
## ?|155060       155060   LOC155060
##                                                          gene_description
## ?|100130426                                                          <NA>
## ?|100133144                                                          <NA>
## ?|100134869 ubiquitin-conjugating enzyme E2Q family member 2 pseudogene 2
## ?|10357                            high mobility group box 1 pseudogene 1
## ?|10431                                                              <NA>
## ?|155060                                              AI894139 pseudogene

One of the advantages of using an ExpressionSet is that the three component objects are always properly paired, and subsetting can be carried out straightforwardly.

tmp <- hnsc[1:100,1:10]
dim(tmp)        # the expression data

## Features  Samples 
##      100       10

dim(Biobase::pData(tmp)) # the phenotypic annotation

## [1] 10  6

dim(Biobase::fData(tmp)) # the genes' annotation

## [1] 100   3

The SummarizedExperiment object

Figure 2: SummarizedExperiment object [source].

The ExpressionSet is generally used for array-based experiments and gene expression data, where the rows are features, and the SummarizedExperiment is generally used for sequencing-based experiments, where the rows are GenomicRanges.

Mapping from ExpressionSet to SummarizedExperiment

It is possible to easily map an ExpressionSet to a SummarizedExperiment.

library(SummarizedExperiment)
sexp <- SummarizedExperiment::makeSummarizedExperimentFromExpressionSet(hnsc)

names(assays(sexp))

## [1] "exprs"

head(SummarizedExperiment::colData(sexp)) # equivalent to pData

## DataFrame with 6 rows and 6 columns
##                 bcr_patient_barcode bcr_sample_barcode tissue_type    my_grade
##                            <factor>           <factor> <character> <character>
## TCGA-CV-6962-11        TCGA-CV-6962    TCGA-CV-6962-11          AN          AN
## TCGA-CV-7432-11        TCGA-CV-7432    TCGA-CV-7432-11          AN          AN
## TCGA-CV-7177-11        TCGA-CV-7177    TCGA-CV-7177-11          AN          AN
## TCGA-CV-7438-11        TCGA-CV-7438    TCGA-CV-7438-11          AN          AN
## TCGA-CV-7440-11        TCGA-CV-7440    TCGA-CV-7440-11          AN          AN
## TCGA-BA-7269-01        TCGA-BA-7269    TCGA-BA-7269-01       Tumor          g1
##                    my_stage patient.anatomic_neoplasm_subdivision
##                 <character>                              <factor>
## TCGA-CV-6962-11          AN                           larynx     
## TCGA-CV-7432-11          AN                           oral cavity
## TCGA-CV-7177-11          AN                           larynx     
## TCGA-CV-7438-11          AN                           oral tongue
## TCGA-CV-7440-11          AN                           larynx     
## TCGA-BA-7269-01   stage iii                           oral tongue

head(SummarizedExperiment::rowData(sexp)) # equivalent to fData

## DataFrame with 6 rows and 3 columns
##             entrez_id gene_symbol
##              <factor>    <factor>
## ?|100130426 100130426   NA       
## ?|100133144 100133144   NA       
## ?|100134869 100134869   UBE2Q2P2 
## ?|10357     10357       HMGB1P1  
## ?|10431     10431       NA       
## ?|155060    155060      LOC155060
##                                                          gene_description
##                                                                  <factor>
## ?|100130426 NA                                                           
## ?|100133144 NA                                                           
## ?|100134869 ubiquitin-conjugating enzyme E2Q family member 2 pseudogene 2
## ?|10357     high mobility group box 1 pseudogene 1                       
## ?|10431     NA                                                           
## ?|155060    AI894139 pseudogene