[BioC] Filter on Fold Change

[Quentin Anstee]

Hi Stephen,

I had read that discussion but misunderstood the conclusions. On re-reading
it, I think you are correct: filtering on fold change in this situation is
bad!

Many thanks,

Quentin 

> -----Original Message-----
> From: Stephen Henderson [mailto:s.henderson at ucl.ac.uk] 
> Sent: 23 February 2006 12:36
> To: Quentin Anstee
> Cc: bioconductor at stat.math.ethz.ch
> Subject: RE: [BioC] Filter on Fold Change
> 
> Hi Quentin
> 
> There has been a similar discussion over the past few days. 
> The main conclusion being (I think) not to filter based upon 
> a known contrast in your data. This will bias any multiple 
> testing corrections you make.
> 
> What you have done so far is OK but if you were going to use 
> something like limma to fit a linear model to your data it 
> would be better to fit it all and select the interesting bits 
> (toptable) based on your contrast afterwards.
> 
> 
> Stephen Henderson
> Wolfson Inst. for Biomedical Research
> Cruciform Bldg., Gower Street
> University College London
> United Kingdom, WC1E 6BT
> +44 (0)207 679 6827
> 
> 
> -----Original Message-----
> From: bioconductor-bounces at stat.math.ethz.ch
> [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of 
> Quentin Anstee
> Sent: 23 February 2006 11:19
> To: bioconductor at stat.math.ethz.ch
> Subject: [BioC] Filter on Fold Change
> 
> Dear List,
>  
> I have used the genefilter package to filter out 
> uninformative probe sets from my GCRMA normalised affy 
> experiment as follows.
>  
> f1 <- kOverA(3,6)
> f2 <- function(x) (IQR(x) > 0.5)
> ff<-filterfun(f1,f2)
> wh<-genefilter(esetGCRMA, ff)
> mySubSet<-esetGCRMA[wh,]
>  
> I would also like to filter out those genes that have less 
> than a 2-fold change in expression between any two of my 
> three study groups *before* I go on to fit a linear model and 
> test for significant differences. My aim is to test as few 
> genes as possible to minimise the effect of multiple testing 
> correction and as I will only follow-up those with at least a 
> 2-fold change, I would like to filter out the rest as soon as 
> possible.
>  
> Please can you advise me whether this can be achieved with genefilter.
> Also,
> any advice on how to script this would also be much 
> appreciated - I have had a look at the vignettes but can't 
> find any that describe filtering on fold change although I 
> see from the list archives that it is commonly done.
>  
> Many thanks,
>  
> Quentin
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> 
> **********************************************************************
> This email and any files transmitted with it are confident...{{dropped}}