[BioC] bug in limma?

Simon Melov smelov at buckinstitute.org
Mon Feb 2 19:02:16 MET 2004

Thanks Gordon...

Actually that wasn't the problem (input data wrong as you suggested). 
The problem was I was reading in more files than I was using. As far as 
I can tell, limma requires the same number of files you initially read 
in to match the numbers of arrays in the target file. I read in over a 
100 arrays, then only described 50 of them in my targets file 
(intending to use the other 50 later). I kept getting the error I 
mentioned, but finally figured out that unless your target file matches 
the number of arrays initially read in, then you will get an error. 
Once I eliminated the excess arrays, they read in fine (i.e. matched 
the target file with the number of arrays read in). Maybe this is my 
ignorance, but this might be worth explicitly mentioning in the guide 
(unless I am misinterpreting my fix).

Any word on the addition of the technical rep and biological rep 
feature you mentioned in your post from September last year? I have 
been treating the technical reps as biological reps, but it would be 
really useful to explicitly handle this via a function within limma.

Thanks again,


More information about the Bioconductor mailing list