# [BioC] limma contrasts

renns renns at reed.edu
Tue Jul 29 20:59:57 CEST 2008

```Thanks for getting back to me Mark.
Actually, I have 12 males and 6 females because there are two different
male phenotypes, T and NT. So I had used:  design_NT1 <-
modelMatrix(targets, ref = "NT1")
MvsF =
(T1+T2+T3+T4+T5+T6+NT2+NT3+NT4+NT5+NT6-F1-F2-F3-F4-F5-F6)/6

but the number of significantly regulated genes that I got was a lot
less than what I expected given the results that I get for TvsF and NTvsF.

I to know what genes are upregulated in all males relative to females
and thought that I could increase my statistical power by considering
MvsF rather than just taking the intersection of the two gene lists that
I got for TvsF and NTvsF. Is that wrong?

Thanks Suzy

Mark Cowley wrote:
> Hi Suzy,
> You have to know which of the labels are from males, or females.
> I'll assume that (a) you have a balanced design and (b) all of the
> even numbers are males, and odd numbers are females, and (c) NT1 is
> the reference, then you'll need to specify a contrast like this:
> MvsF=(T2+T4+T6+F2+F4+F6+NT2+NT4+NT6-T1-T3-T5-F1-F3-F5-NT3-NT5)/9
> cheers,
> Mark
>
> On 25/07/2008, at 5:11 AM, renns wrote:
>
>> I have a loop design experiment in which I compare two types of males
>> and females. There are 6 individuals in each group. For the pairwise
>> comparison I build the contrasts with the following:
>>
>> design_NT1 <- modelMatrix(targets, ref = "NT1")
>> contrasts <- makeContrasts (NvsT =
>> (NT2+NT3+NT4+NT5+NT6-T1-T2-T3-T4-T5-T6)/6, FvsN =
>> (F1+F2+F3+F4+F5+F6-NT2-NT3-NT4-NT5-NT6)/6, TvsF =
>> (T1+T2+T3+T4+T5+T6-F1-F2-F3-F4-F5-F6)/6, levels = design_NT1)
>>
>> Now i want to extract the difference between all males (regardless of
>> type) and females, but I'm not sure how to do this?
>>
>> Thanks
>> Suzy
>>
>> _______________________________________________
>> 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
>
>
>
> -----------------------------------------------------
> Mark Cowley, BSc (Bioinformatics)(Hons)
>
> Peter Wills Bioinformatics Centre
> Garvan Institute of Medical Research, Sydney, Australia
> -----------------------------------------------------
>
>

```