RNA spike-ins: Their applications and bioinformatics methods for the analysis

Question

[Javad]

Hello,

I have a few RNA-seq data which is mixed with ERCC spike-ins. I want to convert the read counts to transcript counts using these ERCC spike-ins. Unfortunately I can not find a good reference which explains how ERCCs work and how they can be used for this type of conversion. I would be grateful if some one can introduce a good reference which explains the application of ERCC spike ins in RNA-seq data analysis and how they can be used for data analysis.

Best

[Devon Ryan]

You don't need ERCC spike-ins for things like TPMs. Unless you have a single-cell dataset, it's best practice to ignore ERCC spike-ins unless you have a really compelling reason to use them (and then only for normalization). Given the issues with them I don't think you can reasonably use them for your purposes (the sequin spike-ins on the other hand...).

[Javad]

Thanks for your response. Actually my data set is a single cell RNA seq data.

[Devon Ryan]

Then you just want to use the spike-ins for library size normalization.

[Carlo Yague]

Hi Devon, out of curiosity, could you elaborate on the issues with the ERCC spike-ins ?

[Devon Ryan]

Different GC distribution and sequence composition bias than real data, different transcript length distribution, not many of them (non-robust normalization), different expression distribution. Have a look at the sequin paper for really well though out spike-ins.