[BioC] edgeR and calcNormFactors manually

[Gordon K Smyth]

Dear Scott,

Normalization scaling factors are specific to each library, so you cannot 
meaningfully compute them for one set of libraries and copy the results to 
a larger set of libraries.

If you try to explain what you are actually trying to achieve, we might be 
able to give you some advice.

As it is, I don't yet understand why you aren't simply using 
calcNormFactors() in the way it is intended, which would be to apply it to 
all libraries at once, or why you need to do any prior ad hoc scaling.

Not sure what you mean by an 'input' sample, because 'input' is usually 
associated with ChIP-seq rather than RNA-seq.

Best wishes
Gordon

> Date: Tue, 15 Apr 2014 16:40:53 -0700
> From: Scott Daniel <scottdaniel at email.arizona.edu>
> To: bioconductor at r-project.org
> Subject: [BioC] edgeR and calcNormFactors manually
>
> Dear Colleagues,
>
> I have an RNA-seq library with 27 samples. 9 of these samples are what we
> call input as the other 18 are derived from them through a biochemical
> process. As such, I have already scaled the genes in those 18 samples to
> roughly add-up to the corresponding inputs.
>
> To manage this at the calcNormFactors step I have the following code:
> y$samples
> y_inputs=y$counts[,grep(".[i]..",colnames(y$counts))]
> norm_factor_inputs <- calcNormFactors(y_inputs)
> y$samples$norm.factors[grep(".[i]..",row.names(y$samples))] =
> norm_factor_inputs
> y$samples$norm.factors[grep(".[p]..",row.names(y$samples))] =
> norm_factor_inputs
> y$samples$norm.factors[grep(".[r]..",row.names(y$samples))] =
> norm_factor_inputs
> y$samples
>
> So what I'm basically trying to do is calcNormFactors for just the inputs
> and then copy them to be the NormFactors for the p and r samples. Is
> copying them like this telling edgeR to re-compute the library sizes? It
> doesn't seem to be happening because the lib.size before and after these
> commands are the same.
>
> Best,
> Scott Daniel
> Zarnescu Lab
> MCB Ph.D. Program
> University of Arizona

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