[BioC] genefilter (nsFilter function)

[Anne]

Hi Robert,


-----Oprindelig meddelelse-----
Fra: Robert Gentleman [mailto:rgentlem at fhcrc.org] 
Sendt: 19. juni 2008 13:46
Til: Anne
Cc: bioconductor at stat.math.ethz.ch
Emne: Re: [BioC] genefilter (nsFilter function)

Hi Anne,

   Perhaps you could tell us the context in which this arose?

Thanks for your answer. 
The code below was used to filter out genes before doing an unsupervised
cluster analysis. 

regards Anne



Anne wrote:
> Hi,
>  
> A question about gene filtering:
>  
> I was wondering if anyone could explain me how I should interpret the 
> code below.
> I am not sure I understand what it means that IQR(x) > 2.5.  
>  
> iqr <- function(x) IQR(x) > 2.5
> filter <- nsFilter(eset, var.filter=TRUE, var.func = iqr, var.cutoff = 
> 0.5)
>  

  The short answer is that for some input, x, the function iqr will return
TRUE of FALSE depending on whether the IQR is larger than 2.5 or smaller
than 2.5 (you could have just tried this).

  > iqr(runif(10))
[1] FALSE

   Somewhat troubling is the use of this function in the call to nsFilter
since the documentation for nsFilter says:

var.func: A 'function' that will be used to assess the variance of a
           probe set across all samples.  This function should return a
           numeric vector of length one when given a numeric vector as
           input.  Probe sets with a 'var.func' value less than
           'var.cutoff' will be removed. The default is 'IQR'.

  and hence the function supplied is not of the correct form, it is unlikely
to raise an error (as there are no simple ways to check whether a function
returns the right kind of value, and logicals can always be coerced to
numeric), but I don't see an easy way to anticipate what nsFilter will do in
this setting.  I think you, or whoever wrote this, would do a bit better to
look at the arguments to the function and use them somewhat more
appropriately (eg, if you set var.cutoff = 2.5 and filterByQuantile=FALSE,
and do not specify a var.func value, I think you get what is intended).


   best wishes
    Robert

>  
>  
> regards
> Anne
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: 
> http://news.gmane.org/gmane.science.biology.informatics.conductor
> 

--
Robert Gentleman, PhD
Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M2-B876
PO Box 19024
Seattle, Washington 98109-1024
206-667-7700
rgentlem at fhcrc.org