How to analyze your RNA sequencing that isn’t just differential expression

Much of the time, a simple differential expression using tools such as EdgeR of DESeq2, is all that is conducted once RNA sequencing is performed. But there are many more insights you can derive from it than that. This article provides an overview of a few of them.

Cell abundance - you may be interested in the cell composition of the microenvironment in your samples. Tools such as xCell and Cibersort can use expression profiles to infer this composition.

Mutation detection- usually you would use DNA sequencing to detect SNPs and mutations but if you don’t have that in your sample you can attempt to detect them in the RNA sequencing. The gene you want to detect mutations in must be expressed, and it is also possible that you can’t detect mutations at the 5’ end of the transcript if you have used polyA sequencing in your library prep.

 Low pass copy number - as with mutation detection , copy number is best performed using DNA sequencing. However, if you dont have that, you can look at the read depth in large windows across each chromosome to call copy number of chromosomes and chromosome arms. Here is a tool that can perform this type of analysis.

Fusion detection - Often fusion detection is much more easily detected in RNA sequencing than in DNA sequencing. There are many tools to do this, with no gold standard. It can throw up a lot of false negatives as well so it i recommended to confirm any of your findings.

Microsatellite instability (MSI) - yes you can also detect MSI from RNAseq! Again, it’s better to do this with DNA sequencing but here is a study where they did this.

Homologous recombination deficiency (HRD) inference- this can be done using gene expression signatures that have been trained using data that are labeled with features of HRD. Tempus performed a large analysis on their RNA samples here.

Pathway signatures - similar to HRD inference, you can create signatures that infer the activity of certain pathways. An example is this MPAS signature that was created to infer MAPK pathway activity. With any of these signatures, be clear on what data was used to train the signature and how your signature may be different.

Advanced QC - Oftentimes you may get results that don’t seem right. Was there a mixup with samples at some point? Is this a contaminated cell line? In these cases you can detect the presence of SNPs in the reads to confirm it is the sample you expected. An example is here for characterizing cell lines using RNAseq data. You can also look for Y chromosome reads in a female sample if the two samples under question are from patients with different genders. You can also use the RNAseq to confirm a mutation you called using a different method.

There’s a lot you can do with RNAseq, so don’t just stick to boring old differential expression!

Come back for next week’s blog post where I will show you novel ways to analyze your exome-seq!