Entering edit mode

Vanessa Vermeirssen
▴
20

@vanessa-vermeirssen-4771
Last seen 7.0 years ago

Dear all,
I need to statistically analyse Nanostring ncounter data to see if
there
is differential expression between experiment and control. I have 3
biological replicates of each and the experimental set-up would
slightly
favor a "_paired_" statistical approach.
Nanostring nCounter data are mRNA counts, like RNA-Seq, but I wonder
if
they have the same properties like RNASeq data i.e. I do only have
the
counts for 110
specifically selected genes. The deeper sampling of one sample
compared
to another e.g. is less
applicable.
The manufacturer suggested some preprocessing of the data: scaling
against positive spike-ins, substracting background (and
absent/present
call generation).
In addition, we performed a normalization with 4 household genes
(selected out of the 8
included in the 110 genes).
I did the DESeq package analysis using these preprocessed data, is
this
package also appropriate in this case (e.g. the library normalization
step?)? Is the preprocessing correct for this?
In addition, I also did a t-test (paired and normal, equal variance,
which I tested, on the log2 data), because this has been described in
literature before.
Another paper describes an FDR permutation approach, but they don't
seem
to have any biological replicates, but 32 control experiments and 10
control genes (Amit et al., 2009).
I also tried to do this on our data.
We have some nice "trends" in our data, which we kind of expected, but
the most significance is obtained with DESeq.
Could you advise me if DESeq is the most correct approach in our case?
What about the other statistical approaches I have tried?
A minor question relates to the preprocessing. How should I deal with
absent/present calls obtained after the preprocessing in the course of
statistical analysis?
Should I include them as NAs from the beginning, or re-evaluate the
results at the end?
Thank you so much in advance already.
Best regards,
Vanessa Vermeirssen
--
=====================================================================
Vanessa Vermeirssen, PhD
Tel:+32 (0)9 331 38 10 Fax:+32 (0)9 3313809
Bioinformatics and Systems Biology
VIB Department of Plant Systems Biology, Ghent University
Technologiepark 927, 9052 Gent, BELGIUM
vamei at psb.vib-ugent.be
http://bioinformatics.psb.ugent.be/

I know this post is ~5 years old, but I found no other reference regarding the background subtraction of Nanostring and deseq2, I got a simple question: Is it right to go this way: ?

regarding the normalization:

Nanostring is : Positive normalization -> Negative norm -> whole codeset normalization (

what is the difference between + norm and whole codeset norm by the way? Is it one to eliminate the hybridization noise and the other to eliminate sample to sample variability?)Deseq2 : whole codeset normalization without background normalization.

But deseq2 needs counts, so is it more accurate if i do the Positive and negative normalization manually according to the manufacturer's guidelines, then convert all the numbers to INTEGERS (down gradeing with excel int() ), and then give it to deseq2? ( After all, after the background normalization we have more accurate counts).

Last question: In case my research includes only 21 samples, is deseq2 relevant at all ? Due to the use of wald test which is parametric and in the case of low sample number it is recommended to use non-parametric.. ?

I really thank you ! I wish you receive this message :)