Differential Expression fro GEO Data

Question

[kalyanimeha]

I am getting an error while downloading GEO matric series file from URL. But URL contains the matrix file.

library(GEOquery)

library(limma)

url <- "ftp://ftp.ncbi.nih.gov/pub/geo/DATA/SeriesMatrix/GSE34747/GSE34747_series_matrix.txt.gz"

filenm <- "data/GSE34747_series_matrix.txt.gz"

if(!file.exists(filenm)) download.file(url, destfile=filenm)

• Error in download.file(url, destfile = filenm) : cannot open destfile 'data/GSE34747_series_matrix.txt.gz', reason 'No such file or directory'

[russhh]

have you forgot to make the data directory?

[egeulgen]

before all do dir.create("data")

[kalyanimeha]

After creating directory previous errors got solved...thanks for helping out. Getting error while calculating ttest.

head(exprs(gse))

GSM854486 GSM854487 GSM854488 GSM854489 GSM854490 GSM854491

GT_44k_23_P100001 -1.06 -0.13 -3.44 -1.06 -0.10 -3.09

GT_44k_23_P100011 -3.26 -2.76 -2.54 -2.72 -2.77 -2.69

GT_44k_23_P100022 -1.80 -5.93 -6.03 -5.48 -6.56 -6.32

GT_44k_23_P100056 -3.43 -3.15 -2.35 -3.04 -3.17 -2.55

GT_44k_23_P100074 2.09 2.97 2.76 2.38 3.20 3.19

GT_44k_23_P100092 -1.88 -2.33 -1.39 -1.92 -2.14 -1.21

exprs(gse) <- log2(exprs(gse))

Warning message:

NaNs produced

boxplot(exprs(gse),outline=FALSE)

pd <- pData(gse)

SampleGroup <- pd$source_name_ch1

library(genefilter)

destats <- rowttests(exprs(gse),fac=SampleGroup)

Error in rowcoltt(x, fac, tstatOnly, 1L) :

Number of groups is 6, but must be >0 and <=2 for 'rowttests'.

[russhh]

the data looks like it's already transformed

Answer 1

[Sean Davis]

I would do something like:

gse = getGEO("GSE34747", destdir="data")[[1]]

Then, you should be able to use limma directly. Note that the data have already been log-transformed, so taking another log will result in the warnings you see above.

[kalyanimeha]

So, how can I find out up and downregulated gene from gse34747 by using limma or any other package?

[Sean Davis]

limma has a great user guide. I would suggest starting there. If you get stuck, you can ask a new question with the details of what you have tried.