Limma: technical replicates in multi-group design
1
0
Entering edit mode
@gordon-smyth
Last seen 1 hour ago
WEHI, Melbourne, Australia
Dear Mike, Wei, Adam, James, I have to agree with Mike. If I understand it correctly, there is no biological replication at all in this experiment. In that situation, there is no way to assess statistical significance relative to biological variation. limma has ways of making use of technical replicates, but there needs to be biological replication as well. Best wishes Gordon > Date: Fri, 26 Mar 2010 09:49:13 +0100 (CET) > From: Mike Walter <michael_walter at="" email.de=""> > To: bioconductor at stat.math.ethz.ch, Adam Kiezun <akiezun at="" gmail.com="">, > Wei Shi <shi at="" wehi.edu.au=""> > Subject: Re: [BioC] Limma: technical replicates in multi-group design > > Dear Wei, > > In my opinion the way you construct the design matrix the technical > replicates are treated as independent samples. The technical replicates > are handled with the block argument in the duplicateCorrelation() > function. As I understand the experimental design, however, there are > only technical but no biological replicates. I have data from the same > type of experiment and told the guys that there is no way of using any > statistical method to analyse their data. Am I right or is it possible > to use limma in such a case? > > Kind regards, > > Mike > -- > > MFT Services > University of T?bingen > Calwerstr. 7 > 72076 T?bingen/GERMANY > > Tel.: +49 7071 29 83210 > Fax. + 49 7071 29 5228 > web: www.mft-services.de > > > -----Urspr?ngliche Nachricht----- > Von: Wei Shi > Gesendet: 25.03.2010 23:37:26 > An: Adam Kiezun [ > Betreff: Re: [BioC] Limma: technical replicates in multi-group design > >> Dear Adam: >> >> You can use the following commands to create a design matrix for >> your experiment: >> >> s <- factor(c("A","A","B","B","C","C")) >> design <- model.matrix(~0+s) >> colnames(design) <- levels(s) >> fit <- lmFit(x,design) >> >> To do a pairwise comparison, you can use the makeConstrasts function: >> >> contr <- makeContrasts(A-B,A-C,B-C,levels=design) >> fit.contr <- eBayes(contrasts.fit(fit,contr)) >> >> Cheers, >> Wei >> >> Adam Kiezun wrote: >>> Hi, >>> How do I write the design matrix to specify technical replicates in >>> multi-group experiment on a single-channel array (data coming from >>> lumi)? >>> >>> I have 3 groups of samples (A, B, C), each group has 2 technical >>> replicates (so I have 6 expression vectors: A1, A2, B1, B2, B3). How >>> do I write the design matrix to tell limma that A1 and A2 etc are >>> technical replicates and that I want to do all pairwise comparisons >>> between groups (ie. AvsB, AvsC, BvsC)? >>> >>> The limma manual covers technical replicates for two-channel arrays, >>> and the multi-group experiments with biological replicates. I'd like >>> to know how to deal with a one-channel arrays, multi-group experiment >>> with technical replicates. >>> >>> regards >>> Adam ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}
limma limma • 1.1k views
ADD COMMENT
0
Entering edit mode
Adam Kiezun ▴ 50
@adam-kiezun-3879
Last seen 9.6 years ago
>?If I understand it correctly, there is no > biological replication at all in this experiment. ?In that situation, there > is no way to assess statistical significance relative to biological > variation. Thanks Gordon, Mike, Wei, James, I'm not sure if it changes the situation but each of the 3 samples/groups comes from RNA pooled from 5 mice. ./adam
ADD COMMENT
0
Entering edit mode
The fact that it is pooled does not help much. I suggest reading (PNAS 2005, "On the utility of pooling biological samples in microarray experiments") http://www.pnas.org/content/102/12/4252.full?sid=a5ba9110-b5fb- 4dca-b721-9cc2615a252f Specifically, I suggest looking at Figure 5, which is rather interesting. Kasper On Fri, Mar 26, 2010 at 7:47 PM, Adam Kiezun <akiezun at="" gmail.com=""> wrote: >>?If I understand it correctly, there is no >> biological replication at all in this experiment. ?In that situation, there >> is no way to assess statistical significance relative to biological >> variation. > > Thanks Gordon, Mike, Wei, James, > I'm not sure if it changes the situation but each of the 3 > samples/groups comes from RNA pooled from 5 mice. > > ./adam > > _______________________________________________ > 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 >
ADD REPLY

Login before adding your answer.

Traffic: 734 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6