Using Bioconductor for Microarray Analysis

Bioconductor has advanced facilities for analysis of microarray platforms including Affymetrix, Illumina, Nimblegen, Agilent, and other one- and two-color technologies.

Bioconductor includes extensive support for analysis of expression arrays, and well-developed support for exon, copy number, SNP, methylation, and other assays.

Major workflows in Bioconductor include pre-processing, quality assessment, differential expression, clustering and classification, gene set enrichment analysis, and genetical genomics.

Bioconductor offers extensive interfaces to community resources, including GEO, ArrayExpress, Biomart, genome browsers, GO, KEGG, and diverse annotation sources.

1 Sample Workflow

The following code illustrates a typical R / Bioconductor session. It uses RMA from the affy package to pre-process Affymetrix arrays, and the limma package for assessing differential expression.

## Load packages
library(affy)   # Affymetrix pre-processing
library(limma)  # two-color pre-processing; differential
                  # expression
                
## import "phenotype" data, describing the experimental design
phenoData <- 
    read.AnnotatedDataFrame(system.file("extdata", "pdata.txt",
    package="arrays"))

## RMA normalization
celfiles <- system.file("extdata", package="arrays")
eset <- justRMA(phenoData=phenoData,
    celfile.path=celfiles)
## Warning: replacing previous import 'AnnotationDbi::tail' by 'utils::tail'
## when loading 'hgfocuscdf'
## Warning: replacing previous import 'AnnotationDbi::head' by 'utils::head'
## when loading 'hgfocuscdf'
## 
## differential expression
combn <- factor(paste(pData(phenoData)[,1],
    pData(phenoData)[,2], sep = "_"))
design <- model.matrix(~combn) # describe model to be fit

fit <- lmFit(eset, design)  # fit each probeset to model
efit <- eBayes(fit)        # empirical Bayes adjustment
topTable(efit, coef=2)      # table of differentially expressed probesets
##                 logFC   AveExpr         t      P.Value    adj.P.Val
## 204582_s_at  3.468416 10.150533  39.03471 1.969915e-14 1.732146e-10
## 211548_s_at -2.325670  7.178610 -22.73165 1.541158e-11 6.775701e-08
## 216598_s_at  1.936306  7.692822  21.73818 2.658881e-11 7.793180e-08
## 211110_s_at  3.157766  7.909391  21.19204 3.625216e-11 7.969130e-08
## 206001_at   -1.590732 12.402722 -18.64398 1.715422e-10 3.016740e-07
## 202409_at    3.274118  6.704989  17.72512 3.156709e-10 4.626157e-07
## 221019_s_at  2.251730  7.104012  16.34552 8.353283e-10 1.049292e-06
## 204688_at    1.813001  7.125307  14.75281 2.834343e-09 3.115297e-06
## 205489_at    1.240713  7.552260  13.62265 7.264649e-09 7.097562e-06
## 209288_s_at -1.226421  7.603917 -13.32681 9.401074e-09 7.784531e-06
##                    B
## 204582_s_at 19.86082
## 211548_s_at 15.88709
## 216598_s_at 15.48223
## 211110_s_at 15.24728
## 206001_at   14.01955
## 202409_at   13.51659
## 221019_s_at 12.69145
## 204688_at   11.61959
## 205489_at   10.76948
## 209288_s_at 10.53327

A top table resulting from a more complete analysis, described in Chapter 7 of Bioconductor Case Studies, is shown below. The table enumerates Affymetrix probes, the log-fold difference between two experimental groups, the average expression across all samples, the t-statistic describing differential expression, the unadjusted and adjusted (controlling for false discovery rate, in this case) significance of the difference, and log-odds ratio. These results can be used in further analysis and annotation.

      ID logFC AveExpr    t  P.Value adj.P.Val     B
636_g_at  1.10    9.20 9.03 4.88e-14  1.23e-10 21.29
39730_at  1.15    9.00 8.59 3.88e-13  4.89e-10 19.34
 1635_at  1.20    7.90 7.34 1.23e-10  1.03e-07 13.91
 1674_at  1.43    5.00 7.05 4.55e-10  2.87e-07 12.67
40504_at  1.18    4.24 6.66 2.57e-09  1.30e-06 11.03
40202_at  1.78    8.62 6.39 8.62e-09  3.63e-06  9.89
37015_at  1.03    4.33 6.24 1.66e-08  6.00e-06  9.27
32434_at  1.68    4.47 5.97 5.38e-08  1.70e-05  8.16
37027_at  1.35    8.44 5.81 1.10e-07  3.08e-05  7.49
37403_at  1.12    5.09 5.48 4.27e-07  1.08e-04  6.21

[ Back to top ]

2 Installation and Use

Follow installation instructions to start using these packages. You can install affy and limma as follows:

## try http:// if https:// URLs are not supported
source("https://bioconductor.org/biocLite.R")
biocLite(c("affy", "limma"))

To install additional packages, such as the annotations associated with the Affymetrix Human Genome U95A 2.0, use

## try http:// if https:// URLs are not supported
source("https://bioconductor.org/biocLite.R")
biocLite("hgu95av2.db")

Package installation is required only once per R installation. View a /packagesfull list of available packages.

To use the affy and limma packages, evaluate the commands

library("affy")
library("limma")

These commands are required once in each R session.

[ Back to top ]

3 Exploring Package Content

Packages have extensive help pages, and include vignettes highlighting common use cases. The help pages and vignettes are available from within R. After loading a package, use syntax like

help(package="limma")
?topTable

to obtain an overview of help on the limma package, and the topTable function, and

browseVignettes(package="limma")

to view vignettes (providing a more comprehensive introduction to package functionality) in the limma package. Use

help.start()

to open a web page containing comprehensive help resources.

[ Back to top ]

4 Pre-Processing Resources

The following provide a brief overview of packages useful for pre-processing. More comprehensive workflows can be found in documentation (available from package descriptions) and in Bioconductor Books and monographs.

4.1 Affymetrix 3’-biased Array

affy, gcrma, affyPLM

  • Require cdf package, probe package and annotation package
  • All these packages are available from Bioconductor via biocLite()

xps

  • Requires installation of ROOT
  • Uses data files from Affymetrix (.CDF, .PGF, .CLF, .CSV) directly

4.2 Affymetrix Exon ST Arrays

oligo

  • Requires a pdInfoPackage built using pdInfoBuilder
  • This package collates cdf, probe, annotation data together
  • These packages are available from Bioconductor via biocLite()
  • Most cases will require a 64-bit computer running Linux and >= 8Gb RAM

exonmap

  • Requires installation of MySQL and Ensembl core database tables
  • Requires specially modified cdf and affy package
  • Requires a 64-bit computer running Linux and >= 8 Gb RAM

xps

  • Requires installation of ROOT
  • Uses data files from Affymetrix (.CDF, .PGF, .CLF, .CSV) directly
  • Will run on conventional desktop computers

4.3 Affymetrix Gene ST Arrays

oligo

  • Requires a pdInfoPackage built using pdInfoBuilder
  • This package collates cdf, probe, annotation data together
  • These packages are available from Bioconductor via biocLite()

xps

  • Requires installation of ROOT
  • Uses data files from Affymetrix (.CDF, .PGF, .CLF, .CSV) directly

4.4 Affymetrix SNP Arrays

oligo

  • Requires a pdInfoPackage built using pdInfoBuilder
  • This package collates cdf, probe, annotation and HapMap data
  • These packages are available from Bioconductor via biocLite()
  • Not yet capable of processing CNV regions in SNP5.0 and SNP6.0

4.5 Affymetrix Tiling Arrays

oligo

  • Requires a pdInfoPackage built using pdInfoBuilder
  • This package collates data from bpmap and cif files

4.6 Nimblegen Arrays

oligo

4.7 Illumina Expression Microarrays

lumi

  • Requires lumi-specific mapping and annotation packages (e.g., lumiHumanAll.db and lumiHumanIDMapping)

beadarray

  • Requires beadarray-specific mapping and annotation packages (e.g., illuminaHumanv1BeadID.db and illuminaHumanV1.db)

[ Back to top ] 

sessionInfo()
## R version 3.4.1 (2017-06-30)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Debian GNU/Linux 9 (stretch)
## 
## Matrix products: default
## BLAS/LAPACK: /usr/lib/libopenblasp-r0.2.19.so
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
##  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
##  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=C             
##  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
##  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
## 
## attached base packages:
## [1] parallel  methods   stats     graphics  grDevices utils     datasets 
## [8] base     
## 
## other attached packages:
## [1] hgfocuscdf_2.18.0   affy_1.54.0         Biobase_2.36.2     
## [4] BiocGenerics_0.22.0 limma_3.32.3        shiny_1.0.3        
## [7] rmarkdown_1.6       knitr_1.16         
## 
## loaded via a namespace (and not attached):
##  [1] Rcpp_0.12.11          compiler_3.4.1        BiocInstaller_1.26.0 
##  [4] questionr_0.6.1       highr_0.6             rmdformats_0.3.3     
##  [7] tools_3.4.1           zlibbioc_1.22.0       bit_1.1-12           
## [10] digest_0.6.12         memoise_1.1.0         tibble_1.3.3         
## [13] evaluate_0.10.1       RSQLite_2.0           preprocessCore_1.38.1
## [16] rlang_0.1.1           DBI_0.7               rstudioapi_0.6       
## [19] yaml_2.1.14           stringr_1.2.0         IRanges_2.10.2       
## [22] S4Vectors_0.14.3      bit64_0.9-7           stats4_3.4.1         
## [25] rprojroot_1.2         R6_2.2.2              AnnotationDbi_1.38.1 
## [28] bookdown_0.4          blob_1.1.0            magrittr_1.5         
## [31] backports_1.1.0       htmltools_0.3.6       mime_0.5             
## [34] xtable_1.8-2          httpuv_1.3.5          stringi_1.1.5        
## [37] miniUI_0.1.1          affyio_1.46.0

[ Back to top ]