Dear All, 

In the post https://support.bioconductor.org/p/69809/#72868 , the author referred to a mirRNA NanoString platform with more than 200 genes when the normalization process was discussed. My NanoString platform contains less than 200 genes for mRNA expression analysis. 

What would be the recommended normalization process and analysis for a NanoString data set with less than 200 genes (172 endogenous and 22 control genes -  mRNA custom panel)?

Is the number of gene a criterium for the choice of process : 

1. NanoString recommended method implemented in package NanoStringNorm  or

2. RNAseq approach implemented in limma as previously outlined (A: Nanostring analysis with limma) 

My understanding is that NanoString produces count data; therefore, it should be subjected to negative binominal distribution regardless. Is this a correct assumption? Should biological coefficient of variation (BCV) be concerned as described in edgeR? 

Thank you for your advice.

Anita