[BioC] Can logFC be used to determine effect size?

[James W. MacDonald]

Hi Jack,

On 8/9/2013 5:02 AM, Jack [guest] wrote:
> Hello,
> I am trying to work out the meaning of logFC in the toptable from limma package. Some sources say that it is the log2 of FC and others say that it is not.

The logFC column is a coefficient from your model (or the numerator from 
a contrast). If you have set things up correctly, it will be the log2 of 
FC between two or more sample types. In other words, if you are doing a 
simple case vs control, and you specify the model correctly and select 
the proper coefficient for your topTable(), then the logFC will be the 
log fold change between the two groups.

And note that log(A/B) = log(A) - log(B), and what you are doing with 
limma is usually taking logged data and computing the difference in 
means. So you are computing the right hand side of that equation, but it 
is identical to the left hand side.

Best,

Jim





>
> Is this value the actual difference in averages of cases and controls (and as such can it be used to determine effect size)?
>
> Regards,
> Jack
>
>   -- output of sessionInfo():
>
> R version 2.15.2 Patched (2012-11-28 r61176)
>
> --
> 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