Skip to content

Anova

2 messages · Luis Silva, Ramon Diaz-Uriarte

#
Hi,

I have a microarray matrix 3000x70 for example (genes x 
experiments). This data has more than 2 classes. I would like 
to make some filtering on the genes. For example, I would like 
to exclude genes with the same mean expression across all 
classes. This is a problem of ANOVA

H0:mu1=mu2=...=muk
H1: at least one is different

First question: Do you think Anova is reasonable for this?
Second question. How can I do this with R?
Clearly, I'll have to make some loop to apply this operation to 
all the genes.

Thanks

Luis

--
A nossa prenda de Natal:  Kit SAPO.ADSL.PT agora a 0 ? e Tr?fego Ilimitado. S? at? 31 de Dezembro. 
Adira j? em http://www.sapo.pt/kitadslgratis
#
Dear Luis,

You might want to take a look at the multtest and genefilter packages (which 
are part of Bioconductor ---www.bioconductor.org---; links to it also from 
R's web page).

Note, however, that no explicit loop would be necessary anyway, since you can 
use, say, 

apply(my.data, 1, my.anova.function)

asuming your data are in a data frame "my.data", where genes are in rows, and 
your "my.anova.function" returns whichever component from an aov or 
summary(aov) you are interested in.

Hope this helps.

Ram?n
On Thursday 12 December 2002 19:45, Luis Silva wrote: