Question

DiffBind custom normalization factor

0

Entering edit mode

C T ▴ 140

@c-t-5858

Last seen 22 months ago

United States

I've searched through the manual and bioconductor support, there doesn't seem to be a way to specify a custom normalization factor in DiffBind when analyzing using DESeq2. I'm wondering if there is a relatively easy "hacky" way to do this? Thanks in advance for your help.

diffbind • 1.2k views

ADD COMMENT • link updated 1 day ago by Abbie • 0 • written 5.0 years ago by C T ▴ 140

score 0 · Answer 1 · 2020-07-10

I'm expecting to add a straightforward way to supply normalization factors explicitly before the next release.

In the meantime, there is a hacky way to do this. When you are using the defaults method=DBA_DESEQ2 and bFullLibrarySize=TRUE, the normalization factors are set as follows:

libsize <- as.numeric(DBA$class['Reads',])
sizeFactors <- libsize/min(libsize)

You can change the values of DBA$class['Reads',] to whatever you like so that the final sizeFactors end up how you want them.

score 0 · Answer 2 · 2020-11-06

0

Entering edit mode

Rory Stark ★ 5.2k

@rory-stark-5741

Last seen 10 weeks ago

Cambridge, UK

The latest version of DiffBind has a new interface function, dba.normalize(), that allows vectors of normalization factors and library sizes (or a matrix of normalization offsets) to be specified instead of using the internal normalization calculations.

ADD COMMENT • link 4.4 years ago Rory Stark ★ 5.2k

0

Entering edit mode

Hello!

Thanks so much for adding dba.normalize and all the documentation in the manual! I'm hoping I can ask a follow-up question here.

I am trying to apply pre-calculated spike-in normalization factors using dba.normalize. I'm not using the spikein option because I don't have easy access to the bams, and instead, I have a table with my sample_id and the normalization factor I'd like to apply. I read through the manual sections on normalization and dba.normalize, but unless I'm missing something obvious (quite possible?), I don't think there's an easy solution for providing pre-calculated norm factors.

So what I did was the following:

dbObj_deseq_spike <- dba.normalize(mycounts, method = DBA_DESEQ2, normalize = DBA_NORM_LIB)
dbObj_deseq_spike$norm$DESeq2$norm.facs <- spike_factors
dbObj_deseq_spike$norm$DESeq2$norm.calc <- "spike in"
dbObj_deseq_spike$norm$DESeq2$norm.method <- "spike in"

And then continue on; however, the MA plot is... very skewed, so I am pretty confident this is NOT the right approach. I should add that I confirmed theyre in the right order before proceeding.

Could you advise on an appropriate way to pass pre-calculated normalization factors from a Drosophila spike-in to DiffBind?

Thanks so much!

ADD REPLY • link 1 day ago Abbie • 0