Hi all
Hope someone can help with this.
We are working on RNA captureSeq experiments where we perform targeted RNAseq on 20 genes of interest (+ probes for the 92 ERCC standards). In the initial phase of the experiment we were evaluating the panel on technical replicates of preclinical samples, which made comparisons between samples quite easy.
In the second phase we are focusing on clinical samples (N=30) with varying amounts of input RNA used for library prep. As RNA contents of our samples are usually too low to measure we normalize based on input volume. We are not specifically interested in DGE analysis as all our samples are from the same disease state but more in a correct normalization method to compare our target genes between our subjects on a per subject basis.
I've read many of the blogpost written here about the pro's/con's of the use of ERCC spikes in DESeq2 and EdgeR for normalization but I was wondering if it was allowed to perform the standard normalization of these methods, as one of the assumptions is that most of the genes present aren't DE. Most of the genes have 0 or very low counts, some genes not of interest have counts because of specific off-target capture and some randomness is introduced because of non-specific off target capture during the protocol. The distribution of the raw counts of captured and uncaptured samples are also quite similar.
I would like to use the ERCC spikes to correct for the input amount of RNA used followed by DESeq2/EgdeR normalization to have a look at our genes of interest between samples and I was wondering if we could get some input from the community on this.
Thanks! JP
What is the question exactly? With only 20 genes you basically must use the spike-ins as that few genes do not really allow to normalize based on the assumption of DESeq2/edgeR that most genes do not change or changes are somewhat symmetric around a median of 0.
Thanks for the comment ATpoint.
I was just wondering if we could use the information/counts that we get on other genes (not part of the target genes) that we get out of the RNAseq data to use for the normalization with DESeq2/EdgeR.
You have RNA-seq or capture? I do not follow.
We've performed totalRNA library prep followed by twist hybridization capture and sequencing. We use this strategy because we are interested in a small subset of genes however we also get information about other genes because of (non-) specific off target capture.
Off-target is nothing I would base anything on...especially not normalization.