[BioC] What RNA-Seq expression estimates to use for plotting to show voom/limma Differential expression

[Gordon K Smyth]

Please keep the discussion on the Bioconductor mailing list.

On Sat, 8 Feb 2014, Heather Estrella wrote:

> Thank you for responding. The CPM in the voom structure is what I have 
> been using. My concern is that I am not confident I can use CPM to 
> compare across genes.

You can't use CPM to compare across genes, but you can use it to display 
differential expression between conditions, which is what I understood you 
to be asking for.

> The key set of questions I am trying to answer are 
> the following:
>
> 1. What is the estimated expression level of gene A and how does that level compare to a predefined set of other genes?

Just take the log2CPM values from voom, and subtract log2(GeneLength/1000) 
from each row, where GeneLength is the effective length of each gene in 
bases.  This will convert the log2CPM values into log2 FPKM.

Best wishes
Gordon

> 2. Is gene A differentially expressed between 2 sample groups?
>
> I can visually represent this in a couple of boxplots or bargraphs, one 
> for each gene, as long as the estimated expression levels are comparable 
> and realistic.
>
> Thanks,
> Heather
>
>
>
>
> -------- Original message --------
> From: Gordon K Smyth <smyth at wehi.EDU.AU>
> Date: 02/07/2014 11:28 PM (GMT-08:00)
> To: Heather Estrella <hestrella at regulusrx.com>
> Cc: Bioconductor mailing list <bioconductor at r-project.org>
> Subject: What RNA-Seq expression estimates to use for plotting to show voom/limma Differential expression
>
>
> Dear Heather,
>
> voom() outputs normalized log-cpm values.  Just plot them.
>
> For example after
>
>   v <- voom(y, design)
>
> the log-cpm values are in v$E.
>
> Best wishes
> Gordon
>
>> Date: Thu, 6 Feb 2014 15:27:59 -0800
>> From: Heather Estrella <hestrella at regulusrx.com>
>> To: "bioconductor at r-project.org" <bioconductor at r-project.org>
>> Subject: [BioC] What RNA-Seq expression estimates to use for plotting
>>        to show voom/limma Differential expression
>>
>> HI Dr. Smyth,
>>
>> I am trying to find the appropriate mRNA expression level estimates to
>> use for generating plots that demonstrate the differential expression
>> statistics calculated from voom/limma. What do you recommend? FPKM plots
>> don't always agree with my voom/limma results.
>>
>> For example, if the differential expression results from limma report a
>> log Fold change of 2 for group A vs. group B, I want a plot (e.g.
>> boxplot, bar graph, etc.) which visualizes this difference and also
>> provides a realistic estimate of the expression level along the axis.
>>
>> If you have other types of visualization (or normalizations) that you
>> prefer for showing estimated expression levels between groups that also
>> demonstrate the limma differential expression trends, please share. I'm
>> open to any ideas/suggestions that you have.
>>
>> Thanks,
>> Heather

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}