[BioC] HELP! lmFit and duplicateCorrelation

Tiandao Li Tiandao.Li at usm.edu
Tue Oct 9 21:52:04 CEST 2007


Dear Jenny,

Thanks for your help. I messed up rows of MA with rows of MA$genes$Row.

I removed the blanks and controls before furthur analysis. I will sort MA 
by MA$genes$ID to get consecutive spots, and also double check the result.

For the difference between methods of ls and robust in lmFit. I looked 
throught online help. 

"ls" for least squares or "robust" for robust regression.

>From MASS package's help, "robust" is a robust regression using an M 
estimator, with or without (?) Tukey's biweight initialized by a specific 
S-estimator.

And "robust" here is a iterated re-weighted least squares (IWLS), how did 
it iterated? Is this the relation between "ls" and "robust" in lmFit?

For me, I am wondering the difference of differentially expressed gene 
lists using methods of "ls" or "robust" in lmFit.

Thanks,

Tiandao


On Tue, 9 Oct 2007, Jenny Drnevich wrote:


>Thank you for your reply. I have another question if you would help me. If
>I sort the MA by MA$genes$ID, the spacing =1, right? spacing has nothing
>to do with Block, Row, or Column from the gpr file, right?

Correct - the spacing is taken from the order in the MA object and 
does not rely on the Block, Row & Column info. Be careful about 
sorting your MA object - as I said before, other spots such as 
buffers and blanks may appear more than 4 times, and may mess up your 
ordered sets of 4. You should always double check to make sure the 
outcome of sorting is what you intended it to be.

Jenny




>Thanks so much,
>
>Tiandao
>
>On Tue, 9 Oct 2007, Jenny Drnevich wrote:
>
>At 10:38 AM 10/9/2007, Tiandao Li wrote:
> > Dear Jenny,
> >
> > I checked the MA$genes, the following is part of the list
> >
> > 1 45Rev2-A03.g
> > ....
> > 25 45Rev2-A03.g
> > ...
> > 49 45Rev2-A03.g
> >
> > 23 rows between 2 replicate spots, so spacing=24?
>
>Yes.
>
>
> > Thansk,
> >
> > Tiandao
> >
> > On Tue, 9 Oct 2007, Jenny Drnevich wrote:
> >
> > At 09:19 AM 10/9/2007, Tiandao Li wrote:
> > >Hello Ido,
> > >
> > >         Block   Row     Column  ID      Name
> > >519     4       2       17      37A-C02.g       P00765
> > >531     4       4       17      37A-C02.g       P00765
> > >513     4       1       17      37A-C02.g       P00765
> > >
> > >For 37A-C02.g, I only selected the first couple of genes from the topTable
> > >list, all 4 replicate spots are printed consecutively in the same column.
> >
> > This is not the same as consecutively spacing in ROWS, which is what
> > the default value of spacing=1 means. Your spacing is probably the
> > number of spots per row on the array. Check MA$genes to see how many
> > rows of the data matrix separate your duplicates.
> >
> > Jenny
> >
> >
> >
> > >Thanks for your reply.
> > >
> > >Tiandao
> > >
> > >On Tue, 9 Oct 2007, Ido M. Tamir wrote:
> > >
> > >On Tuesday 09 October 2007 01:20, Tiandao Li wrote:
> > > > Hello,
> > > >
> > > > I had arrays with 4 replicate spots per gene. I used limma package for
> > > > data analysis.
> > > >
> > > > > targets
> > > >
> > > >          SlideNumber FileName Cy3 Cy5   Name
> > > > Field1           1 13617731 WBM  WC Field1
> > > > Field2           2 13617730 WBM  WC Field2
> > > > Field3           3 13617724  WC WBM Field3
> > > > Field4           4 13617627  WC WBM Field4
> > > > Field5           5 13617626 WBM  WC Field5
> > > >
> > > > After read in data, normalization, I used the following codes for
> > > > within-array replicate spots.
> > > >
> > > > design <- modelMatrix(targets, ref="WC")
> > > > corfit <- duplicateCorrelation(MA, design, ndups=4) # A slow 
> computation!
> > > > fit <- lmFit(MA, design, ndups=4, correlation=corfit$consensus,
> > > > method="ls")
> > > > fit2 <- lmFit(MA, design, ndups=4, correlation=corfit$consensus,
> > > > method="robust")
> > >
> > >Dear Tiandao,
> > >
> > >unless the replicate spots are consecutive you have to give a spacing
> > >argument to indicate what the replicate spots actually are.
> > >If they are randomly spotted on the array, you would have to rearrange
> > >them somehow and then use the appropriate spacing.
> > >
> > >519     4       2       17      37A-C02.g       P00765
> > >531     4       4       17      37A-C02.g       P00765
> > >
> > >doesn't look as if the replicate spots are one after the other.
> > >
> > >HTH best wishes,
> > >ido
> > >
> > >_______________________________________________
> > >Bioconductor mailing list
> > >Bioconductor at 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
> > >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
> > https://stat.ethz.ch/mailman/listinfo/bioconductor
> > Search the archives:
> > http://news.gmane.org/gmane.science.biology.informatics.conductor
>
>Jenny Drnevich, Ph.D.
>
>Functional Genomics Bioinformatics Specialist
>W.M. Keck Center for Comparative and Functional Genomics
>Roy J. Carver Biotechnology Center
>University of Illinois, Urbana-Champaign
>
>330 ERML
>1201 W. Gregory Dr.
>Urbana, IL 61801
>USA
>
>ph: 217-244-7355
>fax: 217-265-5066
>e-mail: drnevich at uiuc.edu

Jenny Drnevich, Ph.D.

Functional Genomics Bioinformatics Specialist
W.M. Keck Center for Comparative and Functional Genomics
Roy J. Carver Biotechnology Center
University of Illinois, Urbana-Champaign

330 ERML
1201 W. Gregory Dr.
Urbana, IL 61801
USA

ph: 217-244-7355
fax: 217-265-5066
e-mail: drnevich at uiuc.edu

_______________________________________________
Bioconductor mailing list
Bioconductor at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor



More information about the Bioconductor mailing list