Entering edit mode
3.3 years ago
Kevin Blighe
88k
Upon request, a quick tutorial for processing the Agilent micro-RNA (miRNA) microarray data of GSE28955.
1, setup (outside R)
The raw TXT files are contained in: https://ftp.ncbi.nlm.nih.gov/geo/series/GSE28nnn/GSE28955/suppl/GSE28955_RAW.tar
- Download this TAR file
- Unpack it [the TAR file]
- Unzip the txt.gz files
- Store these [txt files] in a directory raw/
Then, create a tab-delimited file, targets.txt, that looks like:
FileName desc group
raw/GSM717459.txt AsPC-1 cancer
raw/GSM717460.txt BxPC-3 cancer
raw/GSM717461.txt Capan-1 cancer
raw/GSM717462.txt Capan-2 cancer
raw/GSM717463.txt CFPAC-1 cancer
raw/GSM717464.txt DanG cancer
raw/GSM717465.txt HPAC cancer
raw/GSM717466.txt HPAF-II cancer
raw/GSM717467.txt Hs700T cancer
raw/GSM717468.txt Hs766T cancer
raw/GSM717469.txt HupT3 cancer
raw/GSM717470.txt HupT4 cancer
raw/GSM717471.txt MIAPaCa-2 cancer
raw/GSM717472.txt Panc-1 cancer
raw/GSM717473.txt SU.86.86 cancer
raw/GSM717474.txt SW1990 cancer
raw/GSM717475.txt Ambion normal
raw/GSM717476.txt Biochain-1 normal
raw/GSM717477.txt Biochain-5 normal
raw/GSM717478.txt Clontech normal
raw/GSM717479.txt Pool_slide1 normal
raw/GSM717480.txt Pool_slide2 normal
raw/GSM717481.txt Pool_slide3 normal
2, setup (in R)
targetsfile <- 'targets.txt'
require(limma)
3, import
# read in the data
# readTargets will by default look for the 'FileName' column in the specified file
targetinfo <- readTargets(targetsfile, sep = '\t')
# convert the data to an EListRaw object, which is a data object for single channel data
# specify green.only = TRUE for Agilent
# retain information about background via gIsWellAboveBG
project <- read.maimages(
targetinfo,
source = 'agilent.median',
green.only = TRUE,
other.columns = 'gIsWellAboveBG')
colnames(project) <- sub('raw\\/', '', colnames(project))
4, normalise
# perform background correction on the fluorescent intensities
project.bgcorrect <- backgroundCorrect(project, method = 'normexp')
# normalize the data with the 'quantile' method
project.bgcorrect.norm <- normalizeBetweenArrays(project.bgcorrect,
method = 'quantile')
5, QC filtering
# filter out:
# - control probes
# - probes with no name
# - probes whose signal falls below background in 3 or more samples
Control <- project.bgcorrect.norm$genes$ControlType==1L
NoSymbol <- is.na(project.bgcorrect.norm$genes$GeneName)
IsExpr <- rowSums(project.bgcorrect.norm$other$gIsWellAboveBG > 0) >= 3
project.bgcorrect.norm.filt <- project.bgcorrect.norm[!Control & !NoSymbol & IsExpr, ]
dim(project.bgcorrect.norm)
dim(project.bgcorrect.norm.filt)
# for replicate probes, replace values with the mean
# ID is used to identify the replicates
project.bgcorrect.norm.filt.mean <- avereps(project.bgcorrect.norm.filt,
ID = project.bgcorrect.norm.filt$genes$ProbeName)
dim(project.bgcorrect.norm.filt.mean)
6, check final output
micro RNA annotation
project.bgcorrect.norm.filt.mean$genes)
Row Col ProbeUID ControlType ProbeName GeneName SystematicName
4 1 4 5 0 A_25_P00010390 hsa-miR-30b hsa-miR-30b
5 1 5 7 0 A_25_P00010956 hsa-miR-379 hsa-miR-379
7 1 7 10 0 A_25_P00010912 hsa-miR-634 hsa-miR-634
10 1 11 16 0 A_25_P00010666 hsa-miR-662 hsa-miR-662
16 1 22 31 0 A_25_P00010697 hsa-miR-519e* hsa-miR-519e*
17 1 23 33 0 A_25_P00010752 hsa-miR-32 hsa-miR-32
expression data
project.bgcorrect.norm.filt.mean$E[1:5,1:5]
GSM717459 GSM717460 GSM717461 GSM717462 GSM717463
A_25_P00010390 7.892000 6.715211 6.396424 7.001778 6.975865
A_25_P00010956 4.823109 4.737696 4.836087 4.814941 4.829373
A_25_P00010912 5.078833 5.173138 5.252949 5.195864 5.093929
A_25_P00010666 5.994890 6.508853 6.375115 6.377417 6.506460
A_25_P00010697 5.279065 5.336263 5.215905 5.150749 5.140071
7, simple QC
See some simple code here: build the expression matrix step by step from GEO raw data
For example:
hist(project.bgcorrect.norm.filt.mean$E)
boxplot(project.bgcorrect.norm.filt.mean$E)
require(PCAtools)
p <- pca(
project.bgcorrect.norm.filt.mean$E,
metadata = project.bgcorrect.norm.filt.mean$targets)
biplot(p, colby = 'group', legendPosition = 'bottom')
8, differential expression: Cancer vs Normal
targets <- project.bgcorrect.norm.filt.mean$targets
targets$group <- factor(targets$group, levels = c('normal','cancer'))
expr <- project.bgcorrect.norm.filt.mean$E
# Create the study design
design <- model.matrix(~ 0 + targets$group)
colnames(design) <- c('normal', 'cancer')
# Fit the linear model on the study's data
project.fitmodel <- lmFit(
expr,
design)
# Applying the empirical Bayes method to the fitted values
# Acts as an extra normalisation step and aims to bring the different
# probe-wise variances to common values
project.fitmodel.eBayes <- eBayes(project.fitmodel)
names(project.fitmodel.eBayes)
# Make individual contrasts: cancer vs normal
res <- makeContrasts(res = 'cancer-normal', levels = design)
res.fitmodel <- contrasts.fit(project.fitmodel.eBayes, res)
res.fitmodel.eBayes <- eBayes(res.fitmodel)
toptable <- topTable(
res.fitmodel.eBayes,
adjust = 'BH',
coef = 'res',
number = Inf,
p.value = 1)
head(toptable)
logFC AveExpr t P.Value adj.P.Val B
A_25_P00010700 -3.827306 5.923667 -46.66596 2.575366e-24 1.506589e-21 45.60756
A_25_P00010424 -4.392465 6.047750 -37.79800 3.112054e-22 9.102757e-20 41.00203
A_25_P00010014 -4.394076 6.138899 -33.61272 4.435544e-21 8.149164e-19 38.39282
A_25_P00010110 -5.514454 6.420097 -33.27500 5.572078e-21 8.149164e-19 38.16726
A_25_P00010425 -3.217659 5.653281 -31.37645 2.096846e-20 2.453310e-18 36.85255
A_25_P00010109 -6.444659 6.695490 -29.48774 8.475559e-20 8.263670e-18 35.45955
Kevin
Thanks for the tutorial Kevin. A minor comment, one never needs to do this
expr <- project.bgcorrect.norm.filt.mean$E. Just pass the whole object tolmFit, then the annotation will be preserved in the final table.Thanks Gordon