GSVA example leukemia returns a 'NULL' object

Question

[mbk0asis]

Hello!

While I tested GSVA using the example dataset, I encountered a problem.

Function 'gsva' returned 'NULL', no further analysis could be processed.

I just copied the codes form the vignette and ran it as described.

I pasted the codes under this question.

Could someone help me?

Thank you!

> library(GSVA)
> library(GSEABase)
> library(GSVAdata)
> library(Biobase)
> library(genefilter)
> library(limma)
> library(RColorBrewer)
> 
> data(c2BroadSets)
> c2BroadSets
GeneSetCollection
  names: NAKAMURA_CANCER_MICROENVIRONMENT_UP, NAKAMURA_CANCER_MICROENVIRONMENT_DN, ..., ST_PHOSPHOINOSITIDE_3_KINASE_PATHWAY (3272 total)
  unique identifiers: 5167, 100288400, ..., 57191 (29340 total)
  types in collection:
    geneIdType: EntrezIdentifier (1 total)
    collectionType: BroadCollection (1 total)
> data(leukemia)
> leukemia_eset
ExpressionSet (storageMode: lockedEnvironment)
assayData: 12626 features, 37 samples 
  element names: exprs 
protocolData
  sampleNames: CL2001011101AA.CEL CL2001011102AA.CEL ... CL2001011152AA.CEL (37 total)
  varLabels: ScanDate
  varMetadata: labelDescription
phenoData
  sampleNames: CL2001011101AA.CEL CL2001011102AA.CEL ... CL2001011152AA.CEL (37 total)
  varLabels: subtype
  varMetadata: labelDescription channel
featureData: none
experimentData: use 'experimentData(object)'
Annotation: hgu95a 
> head(pData(leukemia_eset))
                   subtype
CL2001011101AA.CEL     ALL
CL2001011102AA.CEL     ALL
CL2001011104AA.CEL     ALL
CL2001011105AA.CEL     ALL
CL2001011109AA.CEL     ALL
CL2001011110AA.CEL     ALL
> table(leukemia_eset$subtype)

ALL MLL 
 20  17 
> 
> filtered_eset <- nsFilter(leukemia_eset, require.entrez=TRUE, 
+                           remove.dupEntrez=TRUE, var.func=IQR, 
+                           var.filter=TRUE, var.cutoff=0.5, 
+                           filterByQuantile=TRUE, feature.exclude="^AFFX")
> filtered_eset
$eset
ExpressionSet (storageMode: lockedEnvironment)
assayData: 4291 features, 37 samples 
  element names: exprs 
protocolData
  sampleNames: CL2001011101AA.CEL CL2001011102AA.CEL ... CL2001011152AA.CEL (37 total)
  varLabels: ScanDate
  varMetadata: labelDescription
phenoData
  sampleNames: CL2001011101AA.CEL CL2001011102AA.CEL ... CL2001011152AA.CEL (37 total)
  varLabels: subtype
  varMetadata: labelDescription channel
featureData: none
experimentData: use 'experimentData(object)'
Annotation: hgu95a 

$filter.log
$filter.log$numDupsRemoved
[1] 2857

$filter.log$numLowVar
[1] 4292

$filter.log$numRemoved.ENTREZID
[1] 1167

$filter.log$feature.exclude
[1] 19


> 
> leukemia_filtered_eset <- filtered_eset$eset
> leukemia_es <- gsva(leukemia_filtered_eset, c2BroadSets, 
+                     min.sz=10, max.sz=500, verbose=TRUE, 
+                     parallel.sz=1)$es.obs
Mapping identifiers between gene sets and feature names
Estimating GSVA scores for 2019 gene sets.
Computing observed enrichment scores
Estimating ECDFs with Gaussian kernels
Using parallel with 1 cores
  |=========================================================================================================| 100%
> leukemia_es
NULL

Answer 1

[Kevin Blighe]

Some tutorials for this will be out of date because the behaviour of the gsva() function itself changed relatively recently. You now no longer need to access the es.obs object, as it is the only object that is now returned by the gsva() function.

So, you just need:

leukemia_es <- gsva(leukemia_filtered_eset, c2BroadSets, 
  min.sz=10, max.sz=500, verbose=TRUE, 
  parallel.sz=1)

At least I believe this is what is happening here.

Kevin

[mbk0asis]

Thank you, Kevin! That was exactly the problem here!

Apparently, I was following an old tutorial.