[BioC] limma contrast.matrix and design with "common reference"

[James W. MacDonald]

Hi Benedikt,

On 7/25/2014 7:41 AM, Benedikt Athmer [guest] wrote:
> Hi,
>
> I'm confused about creating a contrast matrix with Limma. I've got three groups (LA1777_X, LA_1777_6, LA1777_LVS), each replicted threefold. I would like to use the group LA1777_LVS as a "common reference" to find genes that are expressed differently between LA1777_X and LA1777_6.
>
> The design looks like this:
>
>    LA1777_LVS LA1777_6 LA1777_X
> 1          0        0        1
> 2          0        0        1
> 3          0        0        1
> 4          0        1        0
> 5          0        1        0
> 6          0        1        0
> 7          1        0        0
> 8          1        0        0
> 9          1        0        0
> attr(,"assign")
> [1] 1 1 1
> attr(,"contrasts")
> attr(,"contrasts")$target
> [1] "contr.treatment"
>
> My contrast.matrix looks like this
>
> contrast <- makeContrasts(diff1 = (LA1777_6-LA1777_LVS) - (LA1777_X-LA1777_LVS),
>                            diff2 = LA1777_6-LA1777_X, levels = design)
>> contrast
>              Contrasts
> Levels       diff1 diff2
>    LA1777_LVS     0     0
>    LA1777_6       1     1
>    LA1777_X      -1    -1

The two contrasts you have set up are identical, which is why you get 
identical results. In other words, if you take

(x - y) - (z - y)

this is the same as

x - y - z + y

which simplifies to

x - z

The term 'common reference' is in general used when analyzing two-color 
arrays, where the Cy3 channel is an aliquot of the same reference cDNA. 
In that situation the common reference can be used to help eliminate 
some technical variability, as the input is identical, so any 
differences in fluorescent intensities can be assumed to be technical in 
nature.

But in your case you cannot gain anything by using a different array as 
a common reference.

Best,

Jim


>
> I would expect from the contrast diff1 to find genes that differing between LA1777_6/LA1777_LVS and LA1777_X/LA1777_LVS, but diff2 give the same contrast and result.
> Can you show me how to set up the right design and contrast matrix?
>
> Best regards, Benedikt
>
>
>   -- output of sessionInfo():
>
> R version 3.0.0 Patched (2013-04-04 r62494)
> Platform: x86_64-unknown-linux-gnu (64-bit)
>
> locale:
>   [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C         LC_TIME=C            LC_COLLATE=C         LC_MONETARY=C        LC_MESSAGES=C        LC_PAPER=C           LC_NAME=C
>   [9] LC_ADDRESS=C         LC_TELEPHONE=C       LC_MEASUREMENT=C     LC_IDENTIFICATION=C
>
> attached base packages:
> [1] parallel  stats     graphics  grDevices utils     datasets  methods   base
>
> other attached packages:
>   [1] topGO_2.14.0         SparseM_1.03         GO.db_2.10.1         RSQLite_0.11.4       DBI_0.2-7            AnnotationDbi_1.24.0 graph_1.40.0         gtools_3.3.0
>   [9] MapManITAG2.3_1.0    biomaRt_2.18.0       RColorBrewer_1.0-5   gplots_2.12.1        hopach_2.22.0        cluster_1.14.4       limma_3.18.13        Biobase_2.22.0
> [17] BiocGenerics_0.8.0
>
> loaded via a namespace (and not attached):
>   [1] IRanges_1.20.6     KernSmooth_2.23-10 RCurl_1.95-4.1     XML_3.98-1.1       bitops_1.0-6       caTools_1.16       gdata_2.13.2       grid_3.0.0         lattice_0.20-24
> [10] stats4_3.0.0       tools_3.0.0
>
> --
> Sent via the guest posting facility at bioconductor.org.
>
> _______________________________________________
> Bioconductor mailing list
> 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