How to find out the significant gene from GDS.soft file by using bioconductor

Question

[ches51311]

now I have the microarray data from ncbi website http://www.ncbi.nlm.nih.gov/sites/GDSbrowser?acc=GDS4974

and I want to use the bioconductor code to find the significant gene

and now I can use GEOquery libray to load GDS4974.soft data into R (http://www2.warwick.ac.uk/fac/sci/moac/people/students/peter_cock/r/geo/)

and now I want to normalized this microarray(GDS4974.soft) data, but don't know which bioconductor code I can use to normalize it

can anyone tell me which bioconductor code can I use to normalized it?

and if the next step of normalized is using two sample t test? because after normalized it , I want to using two sample t test

[Devon Ryan]

If it's already on GEO, why not just use GEO2R? You can even see the exact R code that it uses.

[orzech_mag]

If you want to use t test, you can just use tools available at GEO Datasets. At the bottom of GEO Datasets page there is an overlap "Compare 2 sets of samples". Then, you have to assign samples to two groups, which you would like to compare. At the end, choose the type of t test and significance level. As a results you get significantly expressed genes.

[ches51311]

thank you, I find the t test at GEO datasets

[Ming Tommy Tang]

I have two posts related to this problem. Use GEOquery bioconductor package http://crazyhottommy.blogspot.com/search/label/GEO

Answer 1

[Sean Davis]

GEO datasets are already normalized by the submitter. Roughly, the steps are:

library(GEOquery)
gds = getGEO('GDS4974')
eset = GDS2eSet(gds)

Now, you can follow the limma package user guide to do differential expression.

[ches51311]

I'm just use this code to load GDS4974

but it appear these message and not success

Error in validObject(.Object) :
  invalid class "ExpressionSet" object: featureNames differ between assayData and featureData

[Sean Davis]

Sorry you are having problems. What is the output of sessionInfo()?

[orzech_mag]

I think that if you need to perform just a simple t test for differential expression you should simply use tools available @GEO. Otherwise, DE analysis in Bioconductor is not so easy, especially for beginners (believe me, I've tried it not so long time ago). Firstly, you need to normalize raw data, know all about the normalization method, which you would like to use, and so on, then you are able to find your DEGs. Of course, there is limma package manual where everything provided should be clear. Perhaps, GEO tools are enough for your purpose?