I wish to normalize a scRNASeq dataset with respect to ERCC spike-ins, where "for some of the samples, ERCC spike-in RNA was added to the lysis buffer" and was wondering how to do so?

I have seen similar questions posted on the forum, such as the following: Normalization methods with RNA-Seq ERCC spike in? and have checked out the package RUVSeq in the top answer. However, in this package, it seems that spike-ins are taken as transcripts, whereas in the dataset I'm analyzing it seems as though 'spike-ins' are respective of the cell being sequenced rather than a transcript. In addition, in my research I stumbled upon the following tutorial for preprocessing of scRNASeq data: https://f1000research.com/articles/5-2122/v2 where also, as in the RUVSeq package, "ERCC" is grep'd through the rownames to identify ERCC transcripts.

Which leads me to my second question: what is a spike-in if it is not a transcript - and therefore not within the rownames of the dataset and rather within the samples? Please note that I am not from a biological background and, I have read the wikipedia article, along with many others, in my quest to understand what a spike-in is.

In addition, please note that I have a metadata file that states exactly which of the samples have these spike-ins and which dont.

Any advice on how to proceed would be much appreciated.


  • $\begingroup$ Please use the search function. Normalization methods with RNA-Seq ERCC spike in? $\endgroup$
    – story
    Commented Jul 22, 2019 at 16:05
  • $\begingroup$ @story, If you read the full text you will be able to see that I have already referred to that post $\endgroup$
    – h3ab74
    Commented Jul 22, 2019 at 16:52
  • 1
    $\begingroup$ You should treat your ERCC's as entries in your table of "gene counts". You should have counts for them as individual rows (one for each ERCC) in your expression table. The easiest way to do this is to include them in your alignment by appending them to the genomic fasta file (used in the alignment or index generation). $\endgroup$
    – story
    Commented Jul 23, 2019 at 8:20
  • $\begingroup$ I do not have the raw files, just the final count matrix, with metadata identifying which samples/cells had ERCC spike ins added to the lysis buffer.. Is there an alternative way to do what you are referring to for such instances? I have found nothing.. $\endgroup$
    – h3ab74
    Commented Jul 23, 2019 at 15:28
  • 1
    $\begingroup$ You need to get a quantitative readout of the ERCC's. You could try counting them directly (without alignment) from the .fasta/.fastq files. But you need those original files. The whole idea of adding ERCC's is to control for variation in yield... identical amounts should have been added to each sample that contained them; using that information you can then gauge whether there were major differences in concentrations in your final samples. Are you sure that the count table you have doesn't include them? Seems like significant oversight for whoever provided those files. $\endgroup$
    – story
    Commented Jul 24, 2019 at 9:26


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.