[BioC] Array data vs. Next Gen with log 2 Fold Change

john herbert arraystruggles at gmail.com
Thu Jul 14 06:15:09 CEST 2011


Dear Gordan,
Thank you for your explanation. From my simplistic point of view, I
wonder why it is different each side of the division. So to separate
this out.

logFC <- log2(   (y1+0.5) / (lib.size1+0.5)    /
(y2+0.5)*(lib.size2+0.5)   )

On the left side you divide y1 by lib.size but on the right side you
multiply Y2 by lib.size?

To me, each side of the division does not look equivalent?

Is that definitely right? I make the assumption that quantile
normalisation is not needed?

Please explain.

Thanks.

John.

On Wed, Jul 13, 2011 at 11:05 PM, Gordon K Smyth <smyth at wehi.edu.au> wrote:
> Dear John,
>
> The limma equivalent for RNA-Seq is the edgeR package.  However, with no
> replicates, this won't do you much good.  If you only want log2-fold changes
> from your RNA-Seq data, this is easy, although you have to decide what
> you'll do with zero counts.  I suggest, read in your counts into variables
> y1 and y2, then
>
>   lib.size1 <- sum(y1)
>   lib.size2 <- sum(y2)
>   logFC <- log2((y1+0.5)/(lib.size1+0.5)/(y2+0.5)*(lib.size2+0.5))
>
> Best wishes
> Gordon
>
>> Date: Mon, 11 Jul 2011 22:45:17 +0100
>> From: john herbert <arraystruggles at gmail.com>
>> To: bioconductor at r-project.org
>> Subject: [BioC] Array data vs. Next Gen with log 2 Fold Change
>>
>> I have microarray data, which is 2 colour agilent human of 3 technical
>> replicates. Green dye case and Red dye control. I have analysed in Limma,
>> normalising within arrays and between arrays using aQuantile normalisation.
>>
>> I also have some Next gen RNAseq data that has been mapped to the Refseq
>> transcriptome and I have these raw counts. However there are no replicates;
>> only one case and one control.
>>
>> I want to plot how the Log2 Fold change is correlated between the two data
>> sets as they are looking at similar samples.
>>
>> The microarray data is easy as Limma reports log2 fold change but NGS on
>> the other hand does not.
>>
>> What would be the best package/approach to generating a log2 fold change
>> of the next gen counts?
>>
>> I am thinking they should be quantile normalised as the microarray data
>> is????
>
> ______________________________________________________________________
> The information in this email is confidential and intended solely for the
> addressee.
> You must not disclose, forward, print or use it without the permission of
> the sender.
> ______________________________________________________________________
>



More information about the Bioconductor mailing list