[BioC] limma and 3 group design

[Daren Tan]

I have a 12 samples profiled by U133. They are split into 3 groups,
WT.U (Wild-type not transfected), WT.P (Wild-type transfected with
plasmid only), WT.G (Wild-type transfected with plasmid expressing
gene G). The primary aim is to identify differentially expressed genes
affected by gene G. The commented script and partial output is given
below. The venn diagram in pdf format is attached to this email. My
questions are which part of the venn diagram lists only differentially
expressed genes affected by gene G ? The output of topTable shows only
a single column of F, P.Value, adj.P.Val, what do they mean ? At which
step of limma can I extract a list of differentially expressed genes
affected by gene G ?

#WT.U: Wild-type not transfected
#WT.P: Wild-type transfected with plasmid, this is negative control
#WT.G: Wild-type transfected with plasmid expressing gene
R>m.design <- rep(c("WT.U", "WT.P", "WT.G"), each=4)

# sample.dchip.data.xls contains normalized values of U133plus2.0 using dchip
R>m <- read.delim("sample.dchip.data.xls", sep="\t", row.names=1, as.is=T)
R>colnames(m) <- m.design
R>head(m)
               WT.U    WT.U.1    WT.U.2    WT.U.3      WT.P    WT.P.1
  WT.P.2    WT.P.3      WT.G    WT.G.1    WT.G.2    WT.G.3
1007_s_at 10.349609 10.357068 10.471667 10.296924 10.345104 10.314221
10.318749 10.382413 10.359920 10.389718 10.278656 10.251707
1053_at    7.608040  7.634796  7.690094  7.663549  8.258301  8.092113
8.160973  8.068777  8.223478  8.139658  8.087436  8.119274
117_at     6.251763  6.220520  6.340353  6.330107  6.460388  6.287633
6.276603  6.303828  6.315318  6.207221  6.339102  6.398554
121_at     9.515605  9.021253  9.093810  9.170541  9.595528  9.521937
9.662861  9.613140  9.126300  9.240543  9.019469  9.194492
1255_g_at  7.331892  7.015669  7.142806  7.251481  7.208346  7.206259
7.001182  7.281676  6.653735  6.643784  6.621457  6.751383
1294_at    7.761945  7.717853  7.688467  7.773988  7.688812  7.572314
7.730915  7.807970  7.597436  7.602394  7.590732  7.841878

R>design <- model.matrix(~0+factor(m.design, levels=unique(m.design)))
R>colnames(design) <- unique(m.design)

R>fit <- lmFit(m, design)

R>cont.matrix <- makeContrasts("WT.G vs WT.U" =WT.G - WT.U, "WT.G vs
WT.P"=WT.G - WT.P, "WT.P vs WT.U"=(WT.G - WT.P) - (WT.G - WT.U),
levels=unique(m.design))
R>fit2 <- contrasts.fit(fit, cont.matrix)
R>fit2 <- eBayes(fit2)
R>results <- decideTests(fit2)
R>vennDiagram(results, include=c("up", "down"), counts.col=c("red","green"))

R>topTableF(fit2, adjust="BH")
               ID WT.G.vs.WT.U WT.G.vs.WT.P WT.P.vs.WT.U   AveExpr
    F      P.Value    adj.P.Val
31508   222227_at   -2.4414979    2.6066354   5.04813333  7.145136
2245.9382 5.468295e-20 2.989790e-15
15754 206307_s_at    2.5757787    2.3607819  -0.21499676  7.015439
617.1450 1.263117e-15 3.453047e-11
18657   209242_at   -2.2682707   -0.0471402   2.22113050  6.006579
539.2405 3.583991e-15 6.531823e-11
48308   239058_at    1.3057067    1.6484192   0.34271258  6.586163
393.9731 4.020488e-14 5.495505e-10
11885 202436_s_at    1.5394701    1.4423196  -0.09715044  7.286102
323.2023 1.834015e-13 2.005495e-09
40921   231666_at   -1.5600316   -1.7072720  -0.14724040  8.503539
307.0483 2.713968e-13 2.473103e-09
10244   200795_at    1.0378595    0.9476124  -0.09024712  6.258110
291.3060 4.055878e-13 2.782933e-09
33848   224588_at    0.9143286   -0.4761739  -1.39050244 11.188861
290.6110 4.130463e-13 2.782933e-09
32483   223204_at    1.4177460    0.9935906  -0.42415536  6.308256
286.6930 4.580960e-13 2.782933e-09
13807   204359_at    1.3443594    0.7638838  -0.58047559  8.089051
276.7941 5.987744e-13 3.273799e-09
-------------- next part --------------
A non-text attachment was scrubbed...
Name: vennDiagram.pdf
Type: application/pdf
Size: 20018 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20090305/69cadd3d/attachment.pdf>