AffyBatch and ArrayExpress
1
0
Entering edit mode
Ed Siefker ▴ 230
@ed-siefker-5136
Last seen 9 months ago
United States

I'm trying to do DEG analysis on a dataset from ArrayExpress, E-GEOD-38628.   This data is from an Affymetrix 1.0 ST array. The documentation of ArrayExpress() says:

"The output is an object of class ‘AffyBatch’ or ‘ExpressionSet’ or

‘NChannelSet’ with the raw expression values in the assayData of

the object, the information contained in the .sdrf file in the

phenoData, the adf file in the featureData and the idf file

content in the experimentData."

The actual output of ArrayExpress("E-GEOD-38628") is a FeatureSet.

...

E-GEOD-38628  was successfully loaded into  GeneFeatureSet

> mydata.raw

GeneFeatureSet (storageMode: lockedEnvironment)

assayData: 1102500 features, 14 samples


This prevents me from doing GCRMA on this data.

> gcrma(mydata.raw)
Error in (function (classes, fdef, mtable)  :
unable to find an inherited method for function ‘indexProbes’ for signature ‘"GeneFeatureSet", "character"’



Why is ArrayExpress giving me a FeatureSet instead of an AffyBatch?
Can I make ArrayExpress give me an AffyBatch instead?
Should I just use the FeatureSet instead?
What does FeatureSet do that AffyBatch doesn't?
How can I run gcrma() on this data?
What documentation should I have read to understand these issues going in?

Thanks for any insight you can provide.

arrayexpress affy bugs • 673 views
0
Entering edit mode
@james-w-macdonald-5106
Last seen 19 hours ago
United States

The gcrma package is really intended for the old style PM/MM arrays, and is expecting you to be using the affy package instead of oligo. You could probably do some really tricky stuff to get it to work, but in the end I doubt it will really make enough of a difference as compared to just using rma instead.

Put a different way, the Gene ST arrays are PM-only and don't have any MM probes. They do have a set of background probes (the bgp->control->antigenomic probes) that vary from all AT to all GC that you could hypothetically use to estimate GC-specific background binding, but those probes are really artificial, and might not accurately reflect the background binding you should expect. And as I said, it would take some pretty tricky maneuvers to get it done. If I were you I would just use rma on what you have and call it good.

0
Entering edit mode

I know that gcrma requires affybatch.  According to the documentation, ArrayExpress should return an affybatch. What am I missing?

Is gcrma just not recommended anymore?  Is GC bias not an issue with the ST arrays?  How has that been determined?