[BioC] arrayQualityMetrics::aqm.nuse and GeneST

[Wolfgang Huber]

Hi Tim,

this is the result of (perhaps overzealous) tightening of the code that 
I did in the meanwhile. In previous versions, when the computations for 
a certain module (i.e. a figure) of the report failed, the 
'arrayQualityMetrics' function would just silently catch the exception 
and continue with the other modules. Now, the error leads to the 
function being stopped and the error becoming visible to the user. I 
thought that this is more transparent, but it will need further 
refinement to provide an informative error message and concrete 
suggestions of what to do for trouble-shooting. I am looking forward to 
users' feedback here.

In the release version 3.2.0, sadly, not all the consequences of this 
big refactoring exercise have been well enough tested (my fault).

1. To that end, can you make the offending object 'cd8' available to me 
for testing (e.g. via dropbox)?

2. For you to be able to use the package, I propose a two-step solution:
On the short term, copy the function 'arrayQualityMetrics' out of the 
package, and make the part that is headed
"##--Affymetrix specific modules--" optional (e.g. by making it depend 
on a user-definable logical switch).

I will do the same with the devel version, and within 1-2 months, that 
version should again be ready for production use.

Best wishes
	Wolfgang

Tim Rayner scripsit 10/11/10 12:52:
> Hi,
>
> I think I've run into a bug on the latest arrayQualityMetrics release.
> I'm running the function on an AffyBatch object created from HuGene ST
> 1.0 CEL files which I've tested with previous versions of
> arrayQualityMetrics without any problems:
>
>> arrayQualityMetrics(cd8, outdir='canonical_qc')
> The directory 'canonical_qc' has been created.
> Error in quantile.default(dataprep$dat[ss, ], probs = 0.01) :
>    missing values and NaN's not allowed if 'na.rm' is FALSE
> Error in density.default(as.matrix(log2(mm(expressionset))), ...) :
>    'x' contains missing values
> In addition: Warning message:
> In nuse$legend = paste(phrase$fig("Normalized Unscaled Standard Error
> (NUSE)"),  :
>    Coercing LHS to a list
>
> It appears that the aqm.nuse() function is throwing the error. The
> error does not occur if I use CELs from the HG-U133A platform, so I'm
> guessing that it's a problem running the function on CELs from the
> PM-only Gene ST platform?
>
> Aside from that, thank you very much for the improvements with this
> release; the new QC plots very useful indeed.
>
> Best regards,
>
> Tim Rayner
>
>
>
> sessionInfo() follows:
>
> R version 2.12.0 (2010-10-15)
> Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)
>
> locale:
> [1] en_GB.UTF-8/en_GB.UTF-8/C/C/en_GB.UTF-8/en_GB.UTF-8
>
> attached base packages:
> [1] tcltk     stats     graphics  grDevices utils     datasets  methods
> [8] base
>
> other attached packages:
>   [1] Rmpi_0.5-8                affyPara_1.10.0
>   [3] aplpack_1.2.3             snow_0.3-3
>   [5] hugene10stv1cdf_3.24.1    arrayQualityMetrics_3.2.0
>   [7] affyPLM_1.26.0            preprocessCore_1.12.0
>   [9] gcrma_2.22.0              affy_1.28.0
> [11] vsn_3.18.0                Biobase_2.10.0
>
> loaded via a namespace (and not attached):
>   [1] affyio_1.18.0        annotate_1.28.0      AnnotationDbi_1.12.0
>   [4] beadarray_2.0.1      Biostrings_2.18.0    DBI_0.2-5
>   [7] genefilter_1.32.0    grid_2.12.0          hwriter_1.2
> [10] IRanges_1.8.2        lattice_0.19-13      latticeExtra_0.6-14
> [13] limma_3.6.6          marray_1.28.0        RColorBrewer_1.0-2
> [16] RSQLite_0.9-3        simpleaffy_2.26.0    splines_2.12.0
> [19] stats4_2.12.0        survival_2.36-1      SVGAnnotation_0.6-0
> [22] tools_2.12.0         XML_3.2-0            xtable_1.5-6
>
>


-- 


Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber