[BioC] Agilent G4112A Arrays

Chuming Chen chumingchen at gmail.com
Thu Jan 28 05:50:48 CET 2010


Prashantha and all,

Here is the sessional information regarding my analysis of this data set.

Can you point out what I might do wrong?

Thanks,

Chuming

 > library(limma)
 >
 > targets <- readTargets("Targets.txt")
 > targets
  SlideNumber   Name                            FileName Cy3 Cy5
1           1 B1vsT1 US23502303_251239134396_S02_44k.txt  B1  T1
2           2 B2vsT2 US23502303_251239134397_S01_44k.txt  B2  T2
3           3 B3vsT3 US23502303_251239134398_S01_44k.txt  B3  T3
4           4 B4vsT4 US23502303_251239134399_S01_44k.txt  B4  T4
5           5 B5vsT5 US23502303_251239134400_S01_44k.txt  B5  T5
 >
 > RG <- read.maimages(targets, source="agilent")
Read US23502303_251239134396_S02_44k.txt
Read US23502303_251239134397_S01_44k.txt
Read US23502303_251239134398_S01_44k.txt
Read US23502303_251239134399_S01_44k.txt
Read US23502303_251239134400_S01_44k.txt
 >
 > RG <- backgroundCorrect(RG, method="normexp", offset=50)
Green channel
Corrected array 1
Corrected array 2
Corrected array 3
Corrected array 4
Corrected array 5
Red channel
Corrected array 1
Corrected array 2
Corrected array 3
Corrected array 4
Corrected array 5
 >
 > plotDensities(RG)
 >
 > MA <- normalizeBetweenArrays(RG,method="vsn")
Loading required package: vsn
Loading required package: Biobase

Welcome to Bioconductor

  Vignettes contain introductory material. To view, type
  'openVignette()'. To cite Bioconductor, see
  'citation("Biobase")' and for packages 'citation(pkgname)'.

vsn2: 43931 x 10 matrix (1 stratum). Please use 'meanSdPlot' to verify 
the fit.
 >
 > plotDensities(MA)
 >
 > f<-factor(targets$Name)
 > design<-model.matrix(~0+f)
 > design
  fB1vsT1 fB2vsT2 fB3vsT3 fB4vsT4 fB5vsT5
1       1       0       0       0       0
2       0       1       0       0       0
3       0       0       1       0       0
4       0       0       0       1       0
5       0       0       0       0       1
attr(,"assign")
[1] 1 1 1 1 1
attr(,"contrasts")
attr(,"contrasts")$f
[1] "contr.treatment"

 > colnames(design) <- levels(f)
 > design
  B1vsT1 B2vsT2 B3vsT3 B4vsT4 B5vsT5
1      1      0      0      0      0
2      0      1      0      0      0
3      0      0      1      0      0
4      0      0      0      1      0
5      0      0      0      0      1
attr(,"assign")
[1] 1 1 1 1 1
attr(,"contrasts")
attr(,"contrasts")$f
[1] "contr.treatment"

 >
 > fit<-lmFit(MA, design)
 > contrasts.matrix <- makeContrasts(B1vsT1,B2vsT2, B3vsT3, B4vsT4, 
B5vsT5, levels=design)
 >
 > fit2 <- contrasts.fit(fit, contrasts.matrix)
 > fit2 <- eBayes(fit2)
Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = 
stdev.coef.lim) :
  No residual degrees of freedom in linear model fits
 >
 > toptable(fit2)
Error in ebayes(fit, ...) :
  No residual degrees of freedom in linear model fits
 >


Prashantha Hebbar wrote:
> Hi Chuming,
>  
> As per your experimental information, you have replicates. Because, 
> you have samples from same tissue with 2 different region across 
> all patients.  So, you should be able to fit linear model. What I 
> guess, there is something wrong in your analysis steps. So, better to 
> send sessional information to list.
>  
> Regards,
> Prashantha
>
> Prashantha Hebbar Kiradi,
> Dept. of Biotechnology,
> Manipal Life Sciences Center,
> Manipal University,
> Manipal, India
> Email:prashantha.hebbar at manipal.edu
>
> --- On *Mon, 1/25/10, Wolfgang Huber /<whuber at embl.de>/* wrote:
>
>
>     From: Wolfgang Huber <whuber at embl.de>
>     Subject: Re: [BioC] Agilent G4112A Arrays
>     To: "Naomi Altman" <naomi at stat.psu.edu>
>     Cc: "Chuming Chen" <chumingchen at gmail.com>, "Prashantha Hebbar"
>     <prashantha.hebbar at yahoo.com>, bioconductor at stat.math.ethz.ch
>     Date: Monday, January 25, 2010, 8:06 PM
>
>     Hi Chuming
>
>     if you want to work with the approximation that M-values have
>     equal variances, then preprocessing the data with a method that
>     provides variance stabilisation (e.g. vsn) will likely be useful.
>
>     Furthermore, it might be useful to discard a fraction of genes
>     with low A-values, since they are more likely to be either not
>     expressed, or so weakly expressed that you would find it more
>     difficult to validate them.
>
>         Best wishes
>         Wolfgang
>
>     Naomi Altman wrote:
>     > The more data one has, the fewer assumptions one needs.  In the
>     absence of replication, you cannot get p-values without very
>     strong assumptions.  e.g. you could assume that the vast majority
>     of the genes do not differentially express, that their M-values
>     have equal variance and that the M-values are normally
>     distributed.  Then you could use e.g. the IQR of the M-values to
>     estimate the sd and use this to pick a fold cut-off for DE.  You
>     have no reasonable way to estimate FDR with this approach, but it
>     might be slightly better than using 2-fold - or then again, it
>     might not.  Without replication, there is no way to know.
>     >
>     > Regards,
>     > Naomi Altman
>     >
>     >
>     > At 08:53 AM 1/25/2010, Chuming Chen wrote:
>     >> Hi Prashantha,
>     >>
>     >> Thank you for your suggestion. My target file is as below.
>     Although I couldn't fit a linear model, I still wonder whether I
>     can do some statistic on M (log ratio) values and use the p-value
>     to get the differentially expressed genes.
>     >>
>     >> SlideNumber    FileName    Cy3    Cy5
>     >> 1    B1vsT1.txt    B1    T1
>     >> 2    B2vsT2.txt    B2    T2
>     >> 3    B3vsT3.txt    B3    T3
>     >> 4    B4vsT4.txt    B4    T4
>     >> 5    B5vsT5.txt    B5    T5
>     >>
>     >> Chuming
>     >>
>     >>
>     >> Prashantha Hebbar wrote:
>     >>> Dear Chen,
>     >>>
>     >>> You need not to look for any other packages. Since, you do not
>     have any replicates, do not fit linear model, instead just do
>     normalization with in arrays and look at the M (log ratio) values.
>     >>>
>     >>> Regards,
>     >>>
>     >>> Prashantha Hebbar Kiradi,
>     >>> Dept. of Biotechnology,
>     >>> Manipal Life Sciences Center,
>     >>> Manipal University,
>     >>> Manipal, India
>     >>>
>     >>>
>     >>> --- On *Mon, 1/25/10, Chuming Chen /<chumingchen at gmail.com
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=chumingchen@gmail.com>>/*
>     wrote:
>     >>>
>     >>>
>     >>>     From: Chuming Chen <chumingchen at gmail.com
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=chumingchen@gmail.com>>
>     >>>     Subject: [BioC] Agilent G4112A Arrays
>     >>>     To: bioconductor at stat.math.ethz.ch
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=bioconductor@stat.math.ethz.ch>
>     >>>     Date: Monday, January 25, 2010, 6:32 AM
>     >>>
>     >>>     Dear All,
>     >>>
>     >>>     I am trying to find out the differentially expressed genes
>     from
>     >>>     some Agilent Human Whole Genome (G4112A) Arrays data.
>     >>>
>     >>>     I have tried LIMMA package, but LIMMA gave the error
>     message "no
>     >>>     residual degrees of freedom in linear model fits" and
>     stopped. My
>     >>>     guess is that my data has no replicates in the experiment.
>     >>>
>     >>>     Is there any other packages I can use to find differentially
>     >>>     expressed genes which does not require replicates in the
>     experiment?
>     >>>
>     >>>     Thanks for your help.
>     >>>
>     >>>     Chuming
>     >>>
>     >>>     _______________________________________________
>     >>>     Bioconductor mailing list
>     >>>     Bioconductor at stat.math.ethz.ch
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=Bioconductor@stat.math.ethz.ch>
>     >>>     </mc/compose?to=Bioconductor at stat.math.ethz.ch
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=Bioconductor@stat.math.ethz.ch>>
>     >>>     https://stat.ethz.ch/mailman/listinfo/bioconductor
>     >>>     Search the archives:
>     >>> 
>        http://news.gmane.org/gmane.science.biology.informatics.conductor
>     >>>
>     >>
>     >> _______________________________________________
>     >> Bioconductor mailing list
>     >> Bioconductor at stat.math.ethz.ch
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=Bioconductor@stat.math.ethz.ch>
>     >> https://stat.ethz.ch/mailman/listinfo/bioconductor
>     >> Search the archives:
>     http://news.gmane.org/gmane.science.biology.informatics.conductor
>     <http://news..gmane.org/gmane.science.biology.informatics.conductor>
>     >
>     > Naomi S. Altman                                814-865-3791 (voice)
>     > Associate Professor
>     > Dept. of Statistics                              814-863-7114 (fax)
>     > Penn State University                         814-865-1348
>     (Statistics)
>     > University Park, PA 16802-2111
>     >
>     > _______________________________________________
>     > Bioconductor mailing list
>     > Bioconductor at stat.math.ethz.ch
>     <http://us.mc1101.mail.yahoo.com/mc/compose?to=Bioconductor@stat.math.ethz.ch>
>     > https://stat.ethz.ch/mailman/listinfo/bioconductor
>     > Search the archives:
>     http://news.gmane.org/gmane.science.biology..informatics.conductor
>     <http://news.gmane.org/gmane.science.biology.informatics.conductor>
>
>     -- 
>     Best wishes
>          Wolfgang
>
>
>     --
>     Wolfgang Huber
>     EMBL
>     http://www.embl.de/research/units/genome_biology/huber/contact
>
>
>



More information about the Bioconductor mailing list