This is a test version of Biostars. For the public version, visit https://www.biostars.org.
microarray bioinformatic analyses

Hi,

I'm trying to perform microarray differential expression analyses in R using this NCBI dataset: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE159659.

These are the commands I have used so far:

 library(GEOquery)
    library(limma)
    library(preprocessCore)


options("download.file.method.GEOquery" = "wget")

obtain GE data ----------------------------------------------------------

load series (expression matrix) and platform data from GEO

gset <- getGEO("GSE159659", GSEMatrix = TRUE, getGPL= TRUE)
length(gset)
gset <- gset[[1]]

subset data for conditions of interest ----------------------------------

gs0 <- pData(gset)$`subtype:ch1`   
table(gs0)

However for this command table(gs0) it returns the following result:

< table of extent 0 >

It should list the different tissue types (adipose, well differentiated liposarcoma, dedifferentiated liposarcoma) and the quantity of samples belonging to each tissue type.

Does anyone know where I have gone wrong?

Thanks!

microarray
• 2,258 views • link
written by bioinformatics • bioinformatics

1 answer

There is no column subtype:ch1 in your dataset

You need

gs0 <- pData(gset)$`tissue:ch1`   
table(gs0)
gs0
                 adipose tissue    dedifferentiated liposarcoma well differentiated liposarcoma 
                             15                              15                              15
• 0 views • link
written by Basti • Basti

Thanks, it worked. I just had another question.

keep <- c("dedifferentiated liposarcoma", "well differentiated liposarcoma")
sample_idx <- which(gs0 %in% keep)

gset <- gset[, sample_idx]

gs <- pData(gset)$`tissue:ch1`
table(gs)

differential expression analysis ----------------------------------------

assign samples to groups and set up design matrix

gs <- factor(ifelse(gs == "well differentiated liposarcoma", "WDLPS", "DDLPS"))
gset$group <- gs

design <- model.matrix(~group, gset)

For this command design <- model.matrix(~group, gset) I get the following error message: Error in contrasts<-(`tmp`, value = contr.funs[1 + isOF[nn]]) : contrasts can be applied only to factors with 2 or more levels

Do you know I might correct this?

• 0 views • link
written by bioinformatics • bioinformatics

I do not have this error with your code

• 0 views • link
written by Basti • Basti

Apologises I have now retried the code and it worked.

Finally, I tried to run this command below to generate a table of p values for differential expression. However I cannot locate the csv file in the working directory.

Do you know where it might be?

 write.csv(tT, "output/DEA_resul.csv", row.names = FALSE)

The code before this command:

# fit linear model
fit <- lmFit(gset, design)  

# compute statistics
fit2 <- eBayes(fit, 0.01)

# table of DEA results
tT <- topTable(fit2, adjust = "fdr", sort.by = "P", number= Inf)
• 0 views • link
written by bioinformatics • bioinformatics

Check current directory with getwd()

• 0 views • link
written by Basti • Basti

I have checked the desktop and the file is not there. I also get the following error message for write.csv(tT, "output/DEA_resul.csv", row.names = FALSE)

> getwd()
[1] "/Users/mesalie/Desktop"
> write.csv(tT, "output/DEA_resul.csv", row.names = FALSE)

Error in file(file, ifelse(append, "a", "w")) : cannot open the connection In addition: Warning message: In file(file, ifelse(append, "a", "w")) : cannot open file 'output/DEA_resul.csv': No such file or directory

• 0 views • link
written by bioinformatics • bioinformatics

Did you create the folder output ?

• 0 views • link
written by Basti • Basti

No I didn't, thankyou.

Also, after running quality control I received an error message for design <- model.matrix(~group, gset)

pdf("output/1.QC_boxplot.pdf", width = 12, height = 5)
par(mar=c(7,4,2,1))
boxplot(exprs(gset), boxwex = 0.7, notch = TRUE, outline = FALSE, las = 2)
dev.off()

# expression value distribution
pdf("output/2.QC_density_plot.pdf", width = 6, height = 6)
par(mar=c(4,4,2,1))
title <- paste ("GSE159659", "/", annotation(gset), " value distribution", sep ="")
plotDensities(exprs(gset), group = gs, main = title, legend ="topright")
dev.off()

gs <- factor(ifelse(gs == "well differentiated liposarcoma", "WDLPS", "DDLPS"))
gset$group <- gs
design <- model.matrix(~group, gset)

Error in contrasts<-(*tmp*, value = contr.funs[1 + isOF[nn]]) : contrasts can be applied only to factors with 2 or more levels

Do you know how I might correct this?

• 0 views • link
written by bioinformatics • bioinformatics

You changed "well differentiated" to "well-differentiated" which is not present in your dataset. Please check carefully your code before asking such basic question

• 0 views • link
written by Basti • Basti

Log in to answer this question.

More posts like this