[BioC] topGO desperation

[James W. MacDonald]

Hi Christian,

On 5/16/2013 8:14 AM, Oertlin, Christian (Stud. FHML) wrote:
> Hey Jim,
>
> well I guess the problem lies somewhere in how the file is built up. But I cannot find how it should be.
>
> setwd("/Users/christianoertlin/Desktop/Stage files/R")
> read.csv("first.csv", header=TRUE, sep=",")
> library(topGO)
> read.table("first.csv",header=TRUE,sep=",")
> resultFisher<- runTest("first.csv", algrotithm="classic",statistic="Fisher")
> Error in (function (classes, fdef, mtable)  :
>    unable to find an inherited method for function ‘runTest’ for signature ‘"character", "missing", "character"’
>
> This is my problem. The fisher test does not recognize my file as something to work with. I tried moving some rows in the file but it did not work either. Tbh I have no clue what is wrong with it. I added the file in the mail.
>
> The file is an output generated by GO-elite. I already have my GO term with an outcome I just want to do some test and map them into a graph. Like it is done with the from the topGO example.

Well, I can't help you much because I don't know anything about 
GO-elite, and the list strips off most attachments (and I wouldn't have 
opened it anyway. Nothing personal, but I don't even like opening 
attachments from people I know).

A couple of pointers. First, both your read.csv and read.table arguments 
will just spill the input to your screen rather than saving in an 
object. If you want the data to persist, you have to assign to an object 
like thus:

dat <- read.csv("first.csv")

and note that you don't have to repeat default arguments (header is 
already TRUE, and sep is already ",").

I don't know much about topGO, as I have always preferred to use GOstats 
(of course I am being a homer, as I contributed to GOstats, but GOstats 
was written and is maintained by the core group of Bioconductor, so you 
are more likely to get help for that package on this list). I do know 
that the first argument to runTest() isn't a file name, but a topGOdata 
object, so you will have to figure out how to build such a thing first. 
You might want to look here:

http://bioconductor.org/packages/2.12/bioc/vignettes/topGO/inst/doc/topGO.pdf

for help. Or you could take (IMO) the easier route and use GOstats

http://bioconductor.org/packages/2.12/bioc/vignettes/GOstats/inst/doc/GOstatsHyperG.pdf
http://bioconductor.org/packages/2.12/bioc/vignettes/GOstats/inst/doc/GOvis.pdf

but you don't say what the array used was, nor the organism. This might 
make things more difficult. If you are using a less common organism or 
platform, you might want to look here:

http://bioconductor.org/packages/2.12/bioc/vignettes/GOstats/inst/doc/GOstatsForUnsupportedOrganisms.pdf

But you would first do yourself a big favor by reading An Introduction to R

http://cran.r-project.org/doc/manuals/r-release/R-intro.pdf

so you can better familiarize yourself with R first.

Best,

Jim




>
> showSigOfNodes(GOdata, score(resultKS.elim), firstSigNodes = 5, useInfo = 'all')
>
> Kind Regards,
> Christian
>
>
>
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

-- 
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099