[BioC] The units for limma toptable logFC and AveExpr?

[Ying Chen]

Hi guys,

I am new to the limma package. I read the user's guide and it says the logFC and AveExpr are log2 values.

I just tried a test run and found logFC and AveExpr values are really high. For example, following are the values for probe 6733_at (affy HT_HG-U133_Plus_PM custom CDF probe):


            ID       logFC    AveExpr         t      P.Value adj.P.Val         B
14512  6773_at    45.31332   71.14588  27.48457 3.475838e-04 0.5899933 -3.490399

But when I look at the RMAed data before lmFit:

		B02.cel	c03.cel	c04.cel	e04.cel
X6733_at	96.87915039	253.5151215	256.6256409	96.09108734

B02 and e04 are responders, while c03 and c04 are non-responders. For toptable, coef="RESPONDERvsNON_RESPONDER".

RMA was done in aroma.affymetrix because I want to use custom CDF.

>From RMAed data, the ratio or fold change should be less than 3. How come after lmFit, the logFC is about 45?

What did I do wrong?

Thanks a lot for the help!

Ying

> library(aroma.affymetrix)
> ces <- doRMA("Gastric", chipType="HT_HG-U133_Plus_PM,Binary,v14,Hs_ENTREZG", verbose=-5)
> ces
> eset <- extractExpressionSet(ces, verbose=-5)
> library(limma)
> targets <- readTargets("Gastric_Target.txt")
> targets
         Name                       FileName Response
1  GAF-087-P6 5500254086008100810456_B02.CEL      YES
2  GAF-023-P9 5500254086008100810456_C03.CEL       NO
3 GAM-016-P10 5500254086008100810456_C04.CEL       NO
4  GAM-022-P4 5500254086008100810456_E04.CEL      YES
> design <- cbind(NON_RESPONDER=1,RESPONDERvsNON_RESPONDER=targets$Response=="YES")
> design
     NON_RESPONDER RESPONDERvsNON_RESPONDER
[1,]             1                        1
[2,]             1                        0
[3,]             1                        0
[4,]             1                        1
> fit <- lmFit(eset,design)
> fit <- eBayes(fit)
> topTable(fit,coef="RESPONDERvsNON_RESPONDER")
            ID       logFC    AveExpr         t      P.Value adj.P.Val         B
14481  6733_at  -158.58526  175.77775 -76.42590 2.658827e-05 0.5027575 -3.483194
14881  7266_at  -191.56808  498.79973 -55.97116 5.819986e-05 0.5502506 -3.484123
11418 54809_at   126.55865  120.14789  37.65200 1.576701e-04 0.5899933 -3.486542
8501   3429_at  3047.52215 1585.64249  36.09129 1.753662e-04 0.5899933 -3.486931
12470 56623_at   -54.64022   98.54923 -32.38532 2.302226e-04 0.5899933 -3.488091
8143   3156_at  -192.12436  243.72574 -32.15756 2.343398e-04 0.5899933 -3.488176
2949   1286_at   -73.42559   47.14983 -30.94777 2.580292e-04 0.5899933 -3.488657
9626   4259_at -1587.50317 1495.69604 -30.48283 2.680259e-04 0.5899933 -3.488857
5704  23312_at    71.01854  104.03263  28.82746 3.083598e-04 0.5899933 -3.489649
14512  6773_at    45.31332   71.14588  27.48457 3.475838e-04 0.5899933 -3.490399
> write.table(topTable(fit,coef="RESPONDERvsNON_RESPONDER",n=18910),"Gastric_ENTREZG_toptable.txt",sep="\t")
Confidentiality Note:\ This e-mail, and any attachment t...{{dropped:11}}