[BioC] BlueFuse in limmaGUI [was LimmaGUI Error]

[Gordon Smyth]

At 03:19 AM 4/08/2005, Brooke-Powell, Elizabeth wrote:
>Hi Gordon,
>
>Thank you for your help, I am saying that in previous versions of LimmaGUI I
>have not had a problem loading and analyzing BlueFuse data. I load the data
>in following LimmaGUI's series of pop up boxes. First it brings up as box
>asking where the files are, I tell it where the gal file is and then the
>targets file and finally the spot types... it then asks what type of results
>files I have, I click on the other box (because there is no BlueFuse
>option), and then get the box asking what the column headers are for the
>foreground and background for each of the channels. I type in AMPCH2 for the
>Cy5 (foreground and background), then I type in AMPCH1 for Cy3 (fore and
>background) then it goes onto to the pop up box asking if I want to use
>background correction, I select no, and then a pop up window appears for
>spot weighting and If I select yes it tells me that I can only use it when I
>load the data in using the preset formats (in the window where I select
>other) for GenePix and yes all other file types except BlueFuse. If I
>continue it then tells me it will continue without spot weighting... so I
>always select no spot weighting as why see the error... One on the side
>problem is that if for example I want to use GenePix data, but not read in
>the mean data column (that last I knew was preset data column for GenePix
>files), but rather the median data; I need to use the other function in the
>results file type so I tell LimmaGUI which columns to look at and then
>select yes in the spot weighting window it still will not allow me to then
>say which column has flag data so it cannot use spot weight for GenePix data
>in this instance......
>
>I have been using LimmaGUI since it came out and over the last few years
>have had no major problems. Through loading many types of data files and
>comparing them to the ScanArray files I also have I have narrowed down to
>what is causing the error. The problem is not in using BlueFuse data files,
>but rather Fused BlueFuse data files... The fusion combines on slide
>replicates and out puts a different results file (looks the same just only
>has one data point per oligo control etc). I am thinking the problem is
>coming form the fact that the gal file has replicates separated and the
>results files doesn't.

Yes, this is almost certainly the problem. It sounds like you have a gene 
list which is a different length to your expression matrices, and that's a 
fatal problem.

One way for you to go would be to stick to the non-fused data files.

I am meeting with programmers to try to address this problem, in the sense 
that it is not necessary to have a separate GAL file with BlueFuse data. I 
am seeking to "retire" the use of GAL files except with SPOT data. This may 
enable you to work with fused files.

Gordon

>  I am not sure why this matters for plotting an M box
>plot, but do realize it might be a problem later on in analysis (although I
>cannot use the replicate spot function as our replicates are in the same
>blocks so I always keep the reps separate).
>
>Hope that adds further clarification,
>
>Liz