[BioC] multiple groups time course RNA Seq LIMMA

Gordon K Smyth smyth at wehi.EDU.AU
Tue Mar 11 23:20:36 CET 2014

Dear Michela,

I don't really understand what hypothesis you are trying to test, or what 
analysis you have done so far.  If you have already tested for differences 
between treatment and control, what else is it that you want to test? 
I'm not actually even sure what you mean by "treatment", since there is no 
mention of treatment in your target information.

You would need to give reasonably complete code showing what model you've 
fitted in limma, and be more explicit about what you want to test.  It may 
help to read the posting guide:


I think you may have some misconceptions about the use of the 
makeContrasts() function.  It accepts algebraic expressions using the 
column names of the design matrix.  It is not so clever that it can 
understand general English phrases.

Best wishes

Professor Gordon K Smyth,
Bioinformatics Division,
Walter and Eliza Hall Institute of Medical Research,
1G Royal Parade, Parkville, Vic 3052, Australia.

On Tue, 11 Mar 2014, Riba Michela wrote:

> Dear Gordon,

> Thanks for the answer. I've been working on these data for some time. It 
> was obvious how to extract DGE for treatment vs control (just looking at 
> the coefficient names), nevertheless I still do not get how to find 
> genes that change between two experimental groups and the control. I 
> would like to use makeContrasts in order to explicitly define the 
> possibly complex comparisons in my analysis.

> I've tried to write something like:
> contrast.matrix <- makeContrast("(sum of the X.coefficient for group 1) - (sum of the X.coefficient for group2)", levels=design)
> In my first attempt I've set group2 the same as control, this to see if 
> the explicit contrast gives the same results as the standard analysis. 
> Alas, it doesn't.

> How should I specifiy contrasts when more than one factor (i.e. the 
> spline coefficients) characterizes a single condition?

> Best,
> Michela
> Il giorno 15/gen/2014, alle ore 00:53, Gordon K Smyth <smyth at wehi.edu.au> ha scritto:
>> Dear Riba,
>> The advice given in Section 9.6.2 of the User's Guide will still work fine
>> even when Group as more than two levels.
>> Just type colnames(fit) and it will be obvious which coefficients
>> correspond to which experimental group.
>> Yes, you could use contrast.fit() if you wish, but there seems to me to be
>> no reason to do so.
>> Best wishes
>> Gordon
>>> Date: Tue, 14 Jan 2014 11:53:11 +0100
>>> From: Riba Michela <riba.michela at hsr.it>
>>> To: "bioconductor at r-project.org" <bioconductor at r-project.org>
>>> Subject: [BioC] multiple groups time course RNA Seq LIMMA
>>> Hi,
>>> I'm approaching a RNA-seq experiment concerning the analysis of a time
>>> course of 5 time points in 6 experimental groups (including Control
>>> group).
>>> As an example:
>>> FileName Group Time
>>> a Control 6hr
>>> b Control 24h
>>> ...
>>> e ExpG1 6hr
>>> f ExpG1 24hr
>>> ...
>>> l ExpG2 6hr
>>> m ExpG2 24hr
>>> ...
>>> (ExpG1, ExpG2 are experimental groups)
>>> I'm going to use LIMMA for extraction of time changing genes in the
>>> single experimental groups compared to Control group.
>>> I'd like to see how to extract this result in the topTable (i.e. which
>>> coefficients select) for each single comparison of the experimental
>>> group towards Control) since from the provided example in the LIMMA
>>> manual (pag. 49) such topTable is referred to a design concerning one
>>> single experimental group towards Control in a time course instead of
>>> multiple experimental groups.
>>> Is there in addition the possibility to design a contrast matrix in such
>>> situations or is it better to consider topTables using various
>>> coefficients blocks?
>>> Thanks a lot for your answer
>>> Michela Riba
>>> ...
>>> Dr. Michela Riba
>>> Genome Function Unit
>>> Center for Translational Genomics and Bioinformatics
>>> San Raffaele Hospital
>>> Milano
>>> ...
> Dr. Michela Riba
> Genome Function Unit
> Center for Translational Genomics and Bioinformatics
> San Raffaele Scientific Institute
> Via Olgettina 58
> 20132 Milano
> Italy
> lab: +39 02 2643 9114
> skype: mic_mir32
> riba.michela at gmail.com
> riba.michela at hsr.it

The information in this email is confidential and intend...{{dropped:4}}

More information about the Bioconductor mailing list