[BioC] limma genedata

Gordon Smyth smyth at wehi.edu.au
Fri Mar 12 00:12:52 MET 2004


At 06:22 AM 12/03/2004, Dennis Hazelett wrote:
>Hello all,
>Thankyou for your suggestions yesterday (re: getting started), they were 
>extremely helpful. Now I have a more specific question about the limma 
>User's guide. In section 8.2 on the 2-sample w/reference method, the plot 
>commands reference an object "genelist" that points nowhere for me. I have 
>to delet the item to get the command to work, and this causes results to 
>be returned absent identifiers. I'm referring to the following command:
>
>toptable(coef=2,number=15,genelist=genelist[,1:6],fit=fit,eb=eb,adjust="fdr")
>
>? Can I replace
>
>genelist=genelist[,1:6]
>
>with
>genelist=RG$genes
>or something similar?

Yes of course.

But here's two hints. Firstly, in the versions of limma since Christmas, 
limma is now setting the genelist information directly from Genepix gpr 
files. So you don't have to bother with reading the GAL file.

Secondly, you can now use lmFit instead of lm.series, and then simply

topTable(fit)

will find the gene names correctly. See the swirl case study for an example 
of this.

>  It seems to be necessary to get the names of the genes in the output 
> from Bayes analysis. I am using a non-model organism with some cDNA libs 
> of my own construction, so only a fraction of the clones are sequenced 
> and the rest have clone ID's.
>By the way, most of the image links in the LIMMA User's Guide are broken.

This is news to me - I have never heard of this happening before. Please 
advise me off the list the details of your operating system and limma 
version and tell me what's happening exactly.

Gordon

>-Dennis
>p.s. I'm working from a debian linux platform.



More information about the Bioconductor mailing list