Help needed with Nanostring data

[MortG]

NanoStringNorm

I have to work with NanoString data and I wanted to learn this R package called NanoStringNorm but somehow the example codes that are given, do not yield the same results as those displayed. I was wondering if I'm doing something basic very wrong or is their a discrepancy between the given code and the plots in this introductory file.

[kshen]

You can try nanostringnorm, which is a pretty nice R package to do the normalization for nanostring data. (http://cran.r-project.org/web/packag...orm/index.html).

I recently worked on a pretty large nanostring data. There are some technical replicates so I was able to evaluate differential normalization procedures. My current pipeline is to use positive control /negative control /house keeping genes to do the normalization and use DEseq to identify differentially expressed genes.

[kicka11]

No luck so far

Thanks again. Unfortunately no there were no suitable genes in the restricted NanoString set but we are looking back at some similar tissue types which have been previously assessed on the Affymetrix platform.

[frcop]

Housekeeping gene selection

I'm working on plants, so I'm not an expert on these cell types. Based on the names, I can deduce that the genes you list are 'standard' housekeeping genes. While these should work in most cases, this apparently is not the case for you.

The only advice I can give you is to try to select stable genes based on e.g. microarray or RNA-seq data of the cell types (and conditions) you're interested in. Perhaps you're lucky and some or in your nCounter set?

[kicka11]

Housekeeper help

Thanks that's really useful.

The housekeepers we have used seem to be MYC influenced (GAPDH, TUBB, ACTB). Any ideas for genes that may not be so susceptible to MYC (but that would hopefully also have similar expression across B cell lymphoma subtypes)? Am I asking for too much??

[frcop]

Nanostring/nCounter analysis

Nanostring / nCounter data are in the end just counts. So you can use methods that have been developed for RNA-seq counts to analyse these data (e.g. edgeR and DESeq in R/BioConductor). However, you can't use the normalisation procedure of these methods. You need to replace the normalisation values by values you have calculated yourself (first you correct using the positive probes and negative spiked-in probes and next you use normalisation genes included in your probeset). After you replaced the normalisation values, you can proceed with the analysis as described for RNA-seq.

This depends off course on the quality of your normalisation probes, you can select a subset using e.g. geNorm, a tool developed for qPCR analysis.

[kicka11]

I have data on NanoString for samples with two different phenotypes based on gene expression profiling and GSEA: 'high' and 'low'.

A priori we categorised cases as high or low and now tested NanoString on a 53 gene signature for 6 cases (3 previously identified as 'low', 3 previously identified as 'high'; these were tested in duplicate).

I need advice about how I can no analyse the data. We have run normalisations (there are some issues with this) but I cannot work out how to best assess significance and fold change.

Any advice gratefully received.