How to export subset of metadata and expression data from BioConductor GEOquery?

Question

[William]

I am planning to use Bioconductor GEOquery to download a couple of micro-array datasets from NCBI GEO.

Then I would like to export a subset of the metadata and the expression data to flat files that I can import elsewhere.

What I have so far is:

library(GEOquery)
library("R.utils")

geo_id <- "GSE45016"
gse <- getGEO(geo_id,GSEMatrix=FALSE)

#show metadata
Meta(gse)

#show metadata for first sample
GSMList(gse)[[1]]

#select specific field from metadata of first sample
GSMList(gse)[[1]]@header$characteristics_ch1

# Result for sample 1
[1] "tissue: normal prostate (NP) epithelial cells"

GSMList(gse)[[2]]@header$characteristics_ch1

# Result for sample 2
[1] "tissue: prostate cancer cells"   "clinical stage: clinical T4N0M1"
[3] "gleason score: GS 9"             "psa level: PSA 5477ng/ml"

As you can see the number of key value pairs is different for sample 1 and 2. What is would like to have is an array for every key under

@header$characteristics_ch1

and then the value or null (in case the key is missing) for every sample in the GEO dataset" ;

key_tissue: normal prostate (NP) epithelial cells\tprostate cancer cells
key_psa_level: null\tPSA 5477ng/ml

Other metadata fields like "title" luckily only have a single value beneath it.

GSMList(gse)[[1]]@header$title = "Normal prostate"
GSMList(gse)[[2]]@header$title = "High-grade PC1"

Also these I would like to have in an array for the key title.

My second question is how to export the expressions data that is stored under every sample. I would like to stream trough all the probes, get the expression values for that probe for each sample and write it to another csv file.

Answer 1

[Neilfws]

I think that the way you have chosen to read the GSE data into R has created some confusion for you.

Try this instead (note: formatting was lost here so posted as a Gist):

	gse <- getGEO("GSE45016") # you want GSEMatrix = TRUE
	# ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE45nnn/GSE45016/matrix/
	# Found 1 file(s)
	# GSE45016_series_matrix.txt.gz
	# trying URL 'ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE45nnn/GSE45016/matrix/GSE45016_series_matrix.txt.gz'
	# ftp data connection made, file length 1486394 bytes
	# opened URL
	# ==================================================
	# downloaded 1.4 Mb
	#
	# File stored at:
	# /var/folders/7j/7r8lt_3s14s12jx4r2x_qc5c0000gn/T//Rtmpq73dcI/GPL570.soft
	# gse is a list of length 1 named after series matrix; let's tidy it up
	gse <- gse$GSE45016_series_matrix.txt.gz
	gse
	# ExpressionSet (storageMode: lockedEnvironment)
	# assayData: 54675 features, 11 samples
	# element names: exprs
	# protocolData: none
	# phenoData
	# sampleNames: GSM1095876 GSM1095877 ... GSM1095886 (11 total)
	# varLabels: title geo_accession ... data_row_count (34 total)
	# varMetadata: labelDescription
	# featureData
	# featureNames: 1007_s_at 1053_at ... AFFX-TrpnX-M_at (54675 total)
	# fvarLabels: ID GB_ACC ... Gene Ontology Molecular Function (16 total)
	# fvarMetadata: Column Description labelDescription
	# experimentData: use 'experimentData(object)'
	# Annotation: GPL570
	# now get the phenotypic data (covariates etc.) using pData()
	pd <- pData(gse)
	names(pd)
	# [1] "title" "geo_accession" "status" "submission_date" "last_update_date" #
	# [6] "type" "channel_count" "source_name_ch1" "organism_ch1" "characteristics_ch1"
	# [11] "characteristics_ch1.1" "characteristics_ch1.2" "characteristics_ch1.3" "molecule_ch1" "extract_protocol_ch1"
	# [16] "label_ch1" "label_protocol_ch1" "taxid_ch1" "hyb_protocol" "scan_protocol"
	# [21] "description" "data_processing" "platform_id" "contact_name" "contact_email"
	# [26] "contact_department" "contact_institute" "contact_address" "contact_city" "contact_state"
	# [31] "contact_zip/postal_code" "contact_country" "supplementary_file" "data_row_count"
	# sample 1 is normal tissue so does not have cancer-specific data values
	pd$title
	# V2 V3 V4 V5 V6 V7 V8 V9
	# Normal prostate High-grade PC1 High-grade PC2 High-grade PC3 High-grade PC4 High-grade PC5 High-grade PC6 High-grade PC7
	# V10 V11 V12
	# High-grade PC8 High-grade PC9 High-grade PC10
	# 11 Levels: High-grade PC1 High-grade PC10 High-grade PC2 High-grade PC3 High-grade PC4 High-grade PC5 High-grade PC6 ... Normal prostate
	pd$characteristics_ch1
	# V2 V3 V4
	# tissue: normal prostate (NP) epithelial cells tissue: prostate cancer cells tissue: prostate cancer cells
	# V5 V6 V7
	# tissue: prostate cancer cells tissue: prostate cancer cells tissue: prostate cancer cells
	# V8 V9 V10
	# tissue: prostate cancer cells tissue: prostate cancer cells tissue: prostate cancer cells
	# V11 V12
	tissue: prostate cancer cells tissue: prostate cancer cells
	# Levels: tissue: normal prostate (NP) epithelial cells tissue: prostate cancer cells
	pd$characteristics_ch1.1
	# V2 V3 V4 V5
	# clinical stage: clinical T4N0M1 clinical stage: clinical T4N1M1 clinical stage: clinical T2bN1M1
	# V6 V7 V8 V9
	# clinical stage: clinical T3aN0M1 clinical stage: clinical T3bN0M1 clinical stage: clinical T4N1M1 clinical stage: clinical T3bN1M1
	# V10 V11 V12
	# clinical stage: clinical T3bN0M0 clinical stage: clinical T3aN0M0 clinical stage: clinical T3aN1M0
	# 10 Levels: clinical stage: clinical T2bN1M1 clinical stage: clinical T3aN0M0 ... clinical stage: clinical T4N1M1
	pd$characteristics_ch1.2
	# V2 V3 V4 V5 V6 V7 V8
	# gleason score: GS 9 gleason score: GS 9 gleason score: GS 9 gleason score: GS 9 gleason score: GS 9 gleason score: GS 9
	# V9 V10 V11 V12
	# gleason score: GS 8 gleason score: GS 9 gleason score: GS 9 gleason score: GS 8
	# Levels: gleason score: GS 8 gleason score: GS 9
	pd$characteristics_ch1.3
	# V2 V3 V4 V5 V6
	# psa level: PSA 5477ng/ml psa level: PSA 4427ng/ml psa level: PSA1900ng/ml psa level: PSA 630ng/ml
	# V7 V8 V9 V10 V11
	# psa level: PSA 334ng/ml psa level: PSA 311ng/ml psa level: PSA 1000ng/ml psa level: PSA 275ng/ml psa level: PSA 80ng/ml
	# V12
	# psa level: PSA 234ng/ml
	# 11 Levels: psa level: PSA 1000ng/ml psa level: PSA 234ng/ml psa level: PSA 275ng/ml psa level: PSA 311ng/ml ... psa level: PSA1900ng/ml

view raw gistfile1.r hosted with ❤ by GitHub

As for exporting the expression data:

exp <- exprs(gse)

returns a matrix where the column names are sample names.

[Shicheng Guo]

Hi Neilfws, How did you write this reply? first it is in the gitbub and second how to prepare them in gitbub? thanks.

[Sean Davis]

Nicely done.