[BioC] lmFit only deals with numerical values (data matrix object)?

[Sean Davis]

On Fri, Apr 8, 2011 at 9:20 AM, Peter Davidsen <pkdavidsen at gmail.com> wrote:
> Hi Sean,
>
> I read it, but wasn't 100% sure!
>
> Okay, as microarray data need to be log-transformed in order to
> compare gene activities -
> Could I then, if I want to use the lmFit function i limma:
> 1) transform all my log2 ratio data in my data matrix into fold change
> - 2^(log2 ratio)
> 2 divide fold change by 100 => decimal numbers
> 3) add the value +1 to all decimal numbers
> 4) log2 transform all decimal numbers (=> all expression values now numerical)
>
> Sorry for the obvious questions, but I'm new to chip analysis, and
> thus quite unsure

Hi, Peter.

You simply need log ratio data as the input to limma.  There is no
need to perform the steps above if you are starting with log ratios.

I would suggest working through the excellent Limma User Guide very
carefully and be sure to read a little about R along the way.

Sean



> On 8 April 2011 13:28, Sean Davis <sdavis2 at mail.nih.gov> wrote:
>> On Fri, Apr 8, 2011 at 7:13 AM, Peter Davidsen <pkdavidsen at gmail.com> wrote:
>>> Hi List
>>>
>>> Does "lmFit" in limma only deals with numeric values (positive numbers)?
>>> I have created a data matrix object with log2 ratio data (some data
>>> assume negative values).
>>> Can I use my data matrix as object in the lmFit function?
>>
>> Hi, Peter.
>>
>> You should read the help for lmFit, but to answer your question, yes a
>> numeric matrix can be used for lmFit.
>>
>> Sean
>>
>