R help - May 2004 - 1-way anova, nested model

Dear all,

We compared the gene expression level in two conditions: high (H) and low (L),
using real-time PCR. We had 3 samples of each condition, and we quantified each
sample twice (i.e. a technical replication).
Our data looks like: 
H1 H1 H2 H2 H3 H3 L1 L1 L2 L2 L3 L3

We want to test if the level in H vs L is significantly different. I would like
to use ANOVA  and I guess a nested model with the samples as random  variables
should be used. How do you do it with R? (assume we don't have much
experience with R, nor with statistics)


Thanks in advance, Eran
 



Need a new email address that people can remember
Check out the new EudoraMail at
http://www.eudoramail.com

>From what I gather you have 
2 treatments
6 samples nested in treatment
12 replicates in sample.

I do not know what your dependent veriable is, but I would:

code the two treatments as T1 and T2
code the 6 samples as a, b, c, d, e, f so the first three samples
belonging to T1 are not confused with the other three (coding them as
1,2,3 for T1 and 1,2,3 for T2 is a bad idea)
code the replicates as a1, a2, b1, b2....

here I am being extra paranoid in the coding, btw, but it helps and
saves grief later on, in my experience.

The model would be:

library(nlme) # needed!
mod<-lme(response ~ treatment, random = ~ 1|sample/replicate, mydata)

An anova test would give only one F test for the fixed factor, with 1
and 4 df (if I am not mistaken, double check!)

Federico
-- 



================================
Federico C. F. Calboli

Dipartimento di Biologia
Via Selmi 3
40126 Bologna
Italy

tel (+39) 051 209 4187
fax (+39) 051 209 4286

f.calboli at ucl.ac.uk
fcalboli at alma.unibo.it