[BioC] questions of using Limma: should I include all thesam ples?

[Wu, Xiwei]

Fangxin,
Thanks a lot. 
I am also facing a little bit more complicated situation:
There are actually three types of inhibitors, which makes it a 2x4 design
(Am I right?)
If the biological interest is to see each inhibitor effect separately, does
it make sense to include all the chips together for the analysis, or only
one inhibitor at a time?

In addition, what if the inhibitor only experiments are not done? Does it
kill the analysis totally? 
The design matrix would be sth like this:
1 0 0 (only C)
1 1 0 (C+A)
1 1 1 (C+A+B+AB)

to estimate C, A, and B+AB assuming that B effect is little. My question is
whether the linear model still hold (since B is now included in error
variance?). 

Xiwei 

-----Original Message-----
From: Fangxin Hong [mailto:fhong at salk.edu]
Sent: Tuesday, February 08, 2005 8:58 AM
To: Wu, Xiwei
Subject: RE: [BioC] questions of using Limma: should I include all
thesam ples?



> Fangxin,
>
> Thank you very much for your reply.
> Sorry the contrast matrix should read:
> -1  1  0  0
> 1  -1 -1  1
This is right for what you want.


> The design and contrast matrix do look more clear as you suggested, but if
> these different matrix were used, would the result be different at all?
No, there should not be any difference in the result you get.

Fangxin

> Xiwei
>
> -----Original Message-----
> From: Fangxin Hong [mailto:fhong at salk.edu]
> Sent: Monday, February 07, 2005 4:58 PM
> To: Wu, Xiwei
> Cc: 'bioconductor at stat.math.ethz.ch'
> Subject: Re: [BioC] questions of using Limma: should I include all the
> samples?
>
>
>
>> I am trying to use Limma with design matrix of
>>
>>                 1 0 0 0
>>                 1 0 0 0
>>                 1 0 0 0
>>                 0 1 0 0
>>                 0 1 0 0
>>                 0 1 0 0
>>                 0 0 1 0
>>                 0 0 1 0
>>                 0 0 1 0
>>                 0 0 0 1
>>                 0 0 0 1
>>                 0 0 0 1
>>
>> to estimate the four coefficinet of C, C+ A, C+B and C+A+B+AB (of
> course,
>> I
>> can estimate A, B, and AB directly using a different design matrix).
>>
>> Since the contrast of interest is A and AB, so the contrast matrix
> should
>> be:
>> -1  1  0  0
>> -1 -1 -1  1
>>
>> My question is:
>> 1) Are the design and contrast matrix correct?
> If your design matrix is right, then your contrast marix is not right, as
> the (-1,-1,-1,1) will give you estimate of AB-2C, but not AB.
>
> I would suggest you estimate C, A, B, and AB
> using design matrix
> 1 0 0 0 (only C)
> 1 1 0 0(C+A)
> 1 0 1 0(C+B)
> 1 1 1 1 (C+A+B+AB)
>
> and construct your contrast as
> 0 1 0 0 (test A)
> 0 0 0 1  (test AB)
>
>
>
>> 2) I know this is a very naive question, but if I am only interested in
> hormone only effect, can I just use the untreated and hormone alone
> treated
>> samples as the input (so instead of the 12 CEL files, only use the first
> 6
>> CEL files)? Will the analysis result be the same or different if not
> counting the normalization-produced difference? If there is difference, is
>> that due to the difference of df?
> Well, this will only affect your error variance estimation, since you lose
> power for it. Usually less genes will be identified out using subset of
> the data, is indeed you can assume one model for all 12 data sets.
>
> Hopefull this would help.
>
> Fangxin
>
>
>
>
> --
> Fangxin Hong, Ph.D.
> Plant Biology Laboratory
> The Salk Institute
> 10010 N. Torrey Pines Rd.
> La Jolla, CA 92037
> E-mail: fhong at salk.edu
>
>
>
>
>
>
>
> -----------------------------------------------------------
> SECURITY/CONFIDENTIALITY WARNING:  This message and any attachments are
> intended solely for the individual or entity to which they are addressed.
> This communication may contain information that is privileged,
> confidential, or exempt from disclosure under applicable law (e.g.,
> personal health information, research data, financial information).
> Because this e-mail has been sent without encryption, individuals other
> than the intended recipient may be able to view the information, forward
> it to others or tamper with the information without the knowledge or
> consent of the sender. If you are not the intended recipient, or the
> employee or person responsible for delivering the message to the intended
> recipient, any dissemination, distribution or copying of the communication
> is strictly prohibited. If you received the communication in error, please
> notify the sender immediately by replying to this message and deleting the
> message and any accompanying files from your system. If, due t!
>  o the
> security risks, you do not wish to receive further communications via
> e-mail, please reply to this message and inform the sender that you do not
> wish to receive further e-mail from the sender.
> ===========================================================


-- 
Fangxin Hong, Ph.D.
Plant Biology Laboratory
The Salk Institute
10010 N. Torrey Pines Rd.
La Jolla, CA 92037
E-mail: fhong at salk.edu


	[[alternative HTML version deleted]]