[BioC] Re-scaling MA plot Y axis in LimmaGUI?

J.delasHeras@ed.ac.uk J.delasHeras at ed.ac.uk
Thu Dec 1 16:16:07 CET 2005

Hi Keith,

thanks for your reply.

> I'm checking what you say. The resize Window only adjusts the scale of
> the whole plot window, not the scale on either axis.

The resize menu in the latest limmaGUI (July 2005) gives you two 
options: "resize window" (whole plot) and also "resize horizontal (A) 

> There isn't a way of adjusting the scales on the axis in limmaGUI,
> though it is being considered as a future modification.

That would be a very welcome modification, by me at least! :-)

> If an M value is not being plotted, try a different background
> correction method. The Subtract method(the default in limmaGUI) may
> produce some negative values, which are hence excluded. Use None
> instead. There will soon be a change to limmaGUI that includes the
> normexp algorithm which is currently available in limma.

The values that are not shown are a couple of high values. This is in 
an experiment with only two slides at present.
The combined MA plot omits the spot of a gene I am interested in, and 
the M values on the Y axis range between roughly -4 to +4.
The individual MA plot for each slide does show those spots, and the 
scales range between about -7 to 4 (my gene shows M about -6 here) , 
and about -4 to 6 (my gene here shows M about 6).
I used between array normalisation (scale). On the exported toptable, 
the M value is 4.69, which appears to be *just* outside the range 
plotted by Limma. It's as if the spot were discarded, yet it appears at 
the top pof my differentially expressed genes (ranked by B statistic), 
with B=7.55 and P=0.0016 (Holm-adjusted). I am not too worried about 
the P/B values at present as it's just two slides comparing samples 
that are in general quite similar, as I add more replicate experiments 
I am sure the P values will tighten up. But that top gene does appear 
clearly different, and it's got good spot morphology and intensity etc, 
everything seems fine (as far as I can tell)...

so why is it not plotted? (shrug)

> Your right about the command line limma program - more features are
> currently available using it. We plan to add more features to the GUI
> programs. Your comments on which ones (other than settable scale
> parameters) are welcome.

I'm planning to move to limma, but the GUI version is very useful too. 
Personally I would like to see an option to resize the plots, deciding 
what range to plot separately for each axis.

Also, there's one minor niggle that perhaps it can easily be cured? 
When I start a new analysis, I enter a particular spot types file. 
After I see the results, I might want to add further spots, or 
eliminate others, especially for display purposes on MA plots. I can do 
that within LimmaGUI, but I wish those changes could be saved. At 
present it doesn't save them, it sticks to teh original definition for 
the Spot Types.

Having the option to display colour-coded spots on MM plots would also 
be very useful.

More? Okay, one more... ;-)

When I use the Image array plot (to display raw intensities, or ratios, 
or backgrounds, etc), I wish the plot would corresponded to the layout 
of the slide. It doesn't seem to right now. For instance, I am using 
arrays with 24 blocks, arranged in 2 metacolumns and 12 metarows. The 
Image array plot shows all 24 blocks correctly, but arranged in a 4x6 
grid. Which is okay to check each block, but slide-wide patterns in 
background intensities are not as easily seen.

I am very satisfied with LimmaGUI as it is and I think you guys have 
done an excellent job. I hope my suggestions are not taken as 
"complaints" :-) because they're not!

Thanks for all your help,


Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6513374
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR

More information about the Bioconductor mailing list