[BioC] Differential drug effect on clinical groups
James W. MacDonald
jmacdon at uw.edu
Wed Jun 20 18:35:58 CEST 2012
Hi Dave,
On 6/20/2012 12:07 PM, Dave Canvhet wrote:
> Hi James,
>
> This isn't really clear, and I might be way off base with this
> answer, but it looks to me like you are after an interaction term.
> If I were to restate, I would say that you are looking for genes
> that react differently to treatment between the long lived
> integrine positive samples and the short lived integrine negative
> samples.
>
>
> This is exactly what I want, so thanks to your clear restate.
>
> If true, this isn't difficult to set up, although I wouldn't do it
> the way you are. Personally, I would combine the samples into four
> types, based on life and integrine (where for brevity, life is
> long/short and integrine is +/-):
>
> long+
> short+
> long-
> short-
>
> Now your interaction as I understand it will only utilize the
> long+ and short- samples, so you would restrict your samples to
> just those samples that fulfill those criteria. Then you could
> make a lifeinteg factor that is long+ and short- and create a
> design matrix
>
>
> design <- model.matrix(~drug*lifeinteg)
>
>
>
> OK I still to progress on the differences between interaction model
> and additive model (with which I'm more familiar)
> Do you think it will be useful to set up an Intercept ?
> design <- model.matrix(~0+drug*lifeinteg)
There won't be a difference. As an example:
> drug <- factor(rep(1:2, 4))
> lifeinteg <- factor(rep(1:2, each = 4))
> model.matrix(~drug*lifeinteg)
(Intercept) drug2 lifeinteg2 drug2:lifeinteg2
1 1 0 0 0
2 1 1 0 0
3 1 0 0 0
4 1 1 0 0
5 1 0 1 0
6 1 1 1 1
7 1 0 1 0
8 1 1 1 1
attr(,"assign")
[1] 0 1 2 3
attr(,"contrasts")
attr(,"contrasts")$drug
[1] "contr.treatment"
attr(,"contrasts")$lifeinteg
[1] "contr.treatment"
> model.matrix(~0+drug*lifeinteg)
drug1 drug2 lifeinteg2 drug2:lifeinteg2
1 1 0 0 0
2 0 1 0 0
3 1 0 0 0
4 0 1 0 0
5 1 0 1 0
6 0 1 1 1
7 1 0 1 0
8 0 1 1 1
attr(,"assign")
[1] 1 1 2 3
attr(,"contrasts")
attr(,"contrasts")$drug
[1] "contr.treatment"
attr(,"contrasts")$lifeinteg
[1] "contr.treatment"
So the interaction term will be drug2:lifeinteg2 regardless of how you
specify the model.
Best,
Jim
>
> Again many for your time and your help.
>
> Bests
> --
> Dave
>
>
>
>
> and the lifeinteg2 coefficient is the interaction, and gives you
> the genes that react differently to the drug based on being long+
> or short-.
>
> Best,
>
> Jim
>
>
>
>
>
> I've set up my design matrix (target is below):
>
> drug = as.factor(targetATH$drug)
>
> integr = as.factor(targetATH$integrin)
>
> lifetime = as.factor(targetATH$lifetime)
>
> design = model.matrix(~drug+integr+lifetime)
>
> I can't figure out how to set up the correct contrast matrix
> to get the
> coefficient I want.
> I would be very grateful if you could give any pieces of
> advices for that.
> I hope I have enough sample to get enough power to detect some
> genes.
>
>
> many thanks by advance, best regards,
> --
> Dave
>
>
> target :
>
> targetATH
>
> FileName drug lifetime integrin
> 1 sample1.cel Y S +
> 2 sample2.cel Y S +
> 3 sample3.cel Y S +
> 4 sample4.cel Y S +
> 5 sample5.cel Y L +
> 6 sample6.cel Y L +
> 7 sample7.cel Y L +
> 8 sample8.cel Y L +
> 9 sample9.cel Y S -
> 10 sample10.cel Y S -
> 11 sample11.cel Y S -
> 12 sample12.cel Y S -
> 13 sample13.cel Y L -
> 14 sample14.cel Y L -
> 15 sample15.cel Y L -
> 16 sample16.cel Y L -
> 17 sample17.cel N S +
> 18 sample18.cel N S +
> 19 sample19.cel N S +
> 20 sample20.cel N S +
> 21 sample21.cel N L +
> 22 sample22.cel N L +
> 23 sample23.cel N L +
> 24 sample24.cel N L +
> 25 sample25.cel N S -
> 26 sample26.cel N S -
> 27 sample27.cel N S -
> 28 sample28.cel N S -
> 29 sample29.cel N L -
> 30 sample30.cel N L -
> 31 sample31.cel N L -
> 32 sample32.cel N L -
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org <mailto:Bioconductor at r-project.org>
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>
>
> --
> James W. MacDonald, M.S.
> Biostatistician
> University of Washington
> Environmental and Occupational Health Sciences
> 4225 Roosevelt Way NE, # 100
> Seattle WA 98105-6099
>
>
--
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099
More information about the Bioconductor
mailing list