[BioC] Re: ebayes on one array

Gordon Smyth smyth at wehi.edu.au
Thu Aug 14 12:50:01 MEST 2003

At 05:10 AM 14/08/2003, Martin Eklund wrote:
>Dear Dr. Smyth,
>I have a question about LIMMA that I would really appreciate if you could 
>Is it possible to run ebayes() on only one array? The design matrix would 
>then just be a single 1, and this doesnt seem to be appreciated by LIMMA. 
>It gives me the following error message:
>Error in if (out$df.prior == Inf) out$s2.post <- rep(out$s2.prior, 
>length(sigma)) else out$s2.post <- (ifelse(df.residual ==  :
>        missing value where TRUE/FALSE needed
>Is there another way of doing it?

No there isn't. It just doesn't make any sense to run 'ebayes' on one array 
because there aren't any estimated standard deviations to smooth (unless 
you have duplicate spots on the array). If you only have one array, you 
have to rank the genes simply according to their log-ratios.

Would it help to have functionality in 'toptable' to rank genes on 
log-ratio alone when there is only one array?


>Thank you very much for your time!
>Best regards,
>Martin Eklund, MSc student, Uppsala University, Sweden.

More information about the Bioconductor mailing list