Curated breast gene expression data with survival and treatment information

Description

34 manually curated high-quality gene expression microarray datasets with advanced beast cancer samples collected from GEO. All datasets provided have some form of survival and treatment information, and all such clinical variables are semantically normalized across all datasest for easy analyses across datasets. Authors of the Pubmed article linked to each GEO dataset was contacted in an effort to collect as much extra clinical data as possible. See vignette and publication reference from AMIA Translational Science Joint Summits presentation in 2013 for more details on how this data was curated.

Functions are provided to post-process standard S4 ExpressionSet objects to remove samples with high NA rates, impute missing values, collapse duplicated gene symbols or probes, remove duplicated samples that share the same patient ID, and filter genes by variance magnitude or percentile.

Details

Note

Suggestions for new datasets to add are always welcome; the maintainer does aim to only include datasets that have minimal treatment and some form of survival (and/or treatment response) to allow for richer analyses. Raw data is always preferred in order to control normalization schemes. Normalization details for each dataset can be found the the Github repo in the References section.

Author(s)

Katie Planey

Maintainer: Katie Planey <katie.planey@gmail.com>

References

Planey, Butte. Database integration of 4923 publicly-available samples of breast cancer molecular and clinical data. AMIA Joint Summits Translational Science Proceedings. (2003) PMC3814460

Github repo with code, further documentation on datasets and baseline normalization schemes, and database quality checks: https://github.com/kplaney/curatedBreastCancer

Examples

#don't run below in examples() because
#somewhat slow, and similar examples are already run 
#from individual man function files
## Not run: 
#load up master clinical data table
data(clinicalTable)
#Check out the treatment information. 
head(clinicalTable)[,c(112:ncol(clinicalTable))]
#how many had chemotherapy?
numChemoPatients <- length(which(clinicalTable$chemotherapyClass==1))
#how many patients have non-NA OS binary data?


#load up datasets that are in S4 expressionSet format.
#clinical data from master clinicalTable already linked to each sample
#in these ExpressionSets in the phenoData slot.
data(curatedBreastDataExprSetList);

#process only the first two datasets to avoid a long-running example:
#take top 5000 genes by variance from each dataset.
proc_curatedBreastDataExprSetList <- processExpressionSetList(exprSetList=
curatedBreastDataExprSetList[1:2], 
outputFileDirectory = "./", numTopVarGenes=5000)

#now we have processed expression matrices,
#each with the top 5000 genes by variance 

## End(Not run)

Results