Library construction primers containing unique molecular barcode

[hewenjuan]

Hi all,

We are interested in developing our own library construction method for single cell genomics. We want to incorporate unique molecular barcode in the adapters. Does anyone know where to get primers/oligos containing unique molecular barcode synthesized, or do we have to do it by ourselves?

If we have to do it by ourselves, does anyone know a good protocol? It looks very complicated.

Thank you very much in advance!!

[luc]

You could incorporate a stretch of "N"s - lets say eighth of them as UMI. This would be by far the cheapest option.

[hewenjuan]

Thanks, Luc.
You mean when I order oligo, I can just put "N"s there. The synthesized oligo will then contain random barcodes? Will each individual one be unique? As when we analyze data, the same barcode would indicate the same original template.
Thank you!!

[luc]

Yes. That is, what I meant. 10X Genomics uses a 10 nt long UMI for their single-cell kit.

[hewenjuan]

Thanks, Luc.
I contacted IDT regarding synthesis of these primers containing UMI. They told me that " in 1.0 nmole of oligo with an 8 nt UMI, there would be approximately 6.02*10^14 total molecules but only approximately 65,000 unique sequences". If these UMI are not truly unique, then how can we make sure we are not tagging two different original template molecules with the same UMI?

[cmbetts]

Each gene is considered separately in regards to the UMIs and the UMIs will be incorporated in a Poisson distribution (a la digital PCR). In the context of single cells, there's never going to be enough copies of a given transcript to saturate the UMI sequences. Will the efficiency of these protocols and copies per cell, it's even unlikely you'll even get duplicated UMI for a single transcript and considering each one uniquely is good enough unless you actually have enough UMIs seen to approach where Poisson gets away from linear. The trickiest problem is accounting for sequencing errors inflating your counts as you sequence deeply, but there are a few software packages that help with that.

[hewenjuan]

Thanks, cmbetts!