[BioC] CGHcall error

[Daniel Rico]

hits=-2.6 tests=BAYES_00
X-USF-Spam-Flag: NO

Hi Henrik,

Sjoerd Vosse, developer of CGHcall, is having a look at my data.
Thanks to both!

Daniel

Henrik Bengtsson wrote:
> On Feb 5, 2008 3:44 AM, Daniel Rico <drico at cnio.es> wrote:
>   
>> hits=-2.6 tests=BAYES_00
>> X-USF-Spam-Flag: NO
>>
>> Dear List,
>>
>> I am trying to use CGHcall function from CGHcall package, trying to use
>> my own normalized and segmented dataframes (Agilent oligo Human 44A,
>> data normalized with MANOR and segmented with GLAD), buy I get this error:
>>
>> EM algorithm started ...
>> Error en regionsdat[k, 1]:regionsdat[k, 2] : Argumento NA/NaN
>>
>> Which I don't get when I use Wilting data from the vignette example, so
>> it could be a problem with the format of my data (although I can't find
>> any...). I wondered if maybe the dataframes were too large, but I also
>> get (another) error if I only run CGHcall with 1 chromosome:
>>
>> EM algorithm done ...
>> Error en (posteriorfin2[profile == k, ])[, -1] :
>>   número incorreto de dimensiones # Incorrect dimension number
>>     
>
> Without looking at the code itself, that looks like a classical
> mistake.  When writing
>
>   posteriorfin2[profile == k, ]
>
> without an explicit 'drop=FALSE', the developer assumes that 'profile
> == k' will match
> two or more rows in the 'posteriorfin2' matrix/data.frame.  I suspect
> that in your case
> 'profile == k' is only TRUE in one case, which makes
> 'posteriorfin2[profile == k, ]'
> return a vector and not a matrix/data.frame.  This will cause the next
> subsetting '[,-1]'
> to fail, because there are no columns in a plain vector ("Incorrect
> dimension number").
> If the code would have said
>
>  posteriorfin2[profile == k,,drop=FALSE]
>
> the particular error would not show up.
>
> However, in the end of the day, the real question might be why you end
> up with only
> a single case for which 'profile == k' is TRUE.
>
> That's my $0.02
>
> Henrik
>
>
>
>   


-- 
********************************************

Daniel Rico Rodriguez, PhD.
Structural Computational Biology Group
Spanish National Cancer Research Center, CNIO
Melchor Fernandez Almagro, 3. 
28029 Madrid, Spain.
Phone: +34 91 224 69 00 #2256
drico at cnio.es
http://www.cnio.es

********************************************


**NOTA DE CONFIDENCIALIDAD** Este correo electrónico, y ...{{dropped:3}}