john herbert | 12 Oct 09:23 2012
Picon

Re: Limma; a kind of extended paired analyses with or without treatment

Thanks James,
appreciated as you have saved me a lot of time.

John.

On Thu, Oct 11, 2012 at 7:48 PM, James W. MacDonald <jmacdon@...> wrote:
> Ugh. Jumped the gun. This does *not* require you to fit a random effects
> model, as you have done every treatment to cells from each patient. You can
> just block on Sample and then make your comparisons.
>
> In other words, if you add Sample to your design matrix, you will in effect
> be removing the patient-specific effect. Something like
>
> design <- model.matrix(~0+treatment*time+sample)
>
> Best,
>
> Jim
>
>
>
>
> On 10/11/2012 2:27 PM, john herbert wrote:
>>
>> Thanks James,
>> This does not have time course but judging by your answer, I can just
>> add this in, in place of, say, tissue.
>>
>> Kind regards,
>>
>> John.
>>
>> On Thu, Oct 11, 2012 at 7:23 PM, James W. MacDonald<jmacdon@...>
>> wrote:
>>>
>>> Hi John,
>>>
>>>
>>> On 10/11/2012 2:15 PM, john herbert wrote:
>>>>
>>>> Dear all.
>>>> I have been pondering about constructing a design matrix based on the
>>>> Limma user guide, where I combine a time course with a paired
>>>> analyses. The targets file looks like;
>>>>
>>>> Sample  treatment       time
>>>> 1       control 24
>>>> 1       control 72
>>>> 1       control 0
>>>> 1       treatment       24
>>>> 1       treatment       72
>>>> 2       control 24
>>>> 2       control 72
>>>> 2       control 0
>>>> 2       treatment       24
>>>> 2       treatment       72
>>>> 3       control 24
>>>> 3       control 72
>>>> 3       control 0
>>>> 3       treatment       24
>>>> 3       treatment       72
>>>>
>>>> Sample number refers to an individuals cancer cells, treatment refers
>>>> to added drug or not and numbers are in hours (time elapsed). So it is
>>>> a kind of paired, as patient variability is to be considered. The
>>>> control sample at 0 is the same as treatment at time 0 as these are
>>>> the same cells without any time/treatment.
>>>>
>>>> Please could someone help me understand how I can construct a design
>>>> matrix and to understand how I can extract differently expressed genes
>>>> that come about due to time, due to treatment and interaction of them
>>>> both.
>>>>
>>>> Any pointers appreciated, though I am trying to see if the examples in
>>>> the manual can be applied to this scenario.
>>>
>>>
>>> See the multi-level experiment example in the user guide, starting on p.
>>> 47.
>>>
>>> Best,
>>>
>>> Jim
>>>
>>>> Thank you.
>>>>
>>>> John.
>>>>
>>>> _______________________________________________
>>>> Bioconductor mailing list
>>>> Bioconductor@...
>>>> 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
>>>
>
> --
> James W. MacDonald, M.S.
> Biostatistician
> University of Washington
> Environmental and Occupational Health Sciences
> 4225 Roosevelt Way NE, # 100
> Seattle WA 98105-6099
>

_______________________________________________
Bioconductor mailing list
Bioconductor@...
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor


Gmane