Single-cell RNA-seq with ERCC RNA Spike-In

[kobeho24]

Hi all,
I recently came across some single-cell rna-seq experiment, in which people always recommand using ERCC spike-ins for normalization. And I just wonder how much spike-ins should be added per cell, 1ul 1:5,000,000 diluted from original stock? And also how you guys keep these easy-degraded synthetic RNA on your hands? As super diluted RNA is not that stable even in -80C, I am afraid Also, there are two options, the regular one and the Exfold one, which one you guys prefer?

Wish you guys a happy new year

Gary

[ostolop]

Hi!

This is a bit off-topic, and somewhat blatant self-advertising, but for what it's worth, we developed a method -- preparing it for publication, draft going on biorxiv soon -- to deal with single cell RNA-seq without ERCC spike-ins.

Here's some info on it http://genestack.com/single-cell-rnaseq/ and a bit more here https://genestack.com/blog/2014/09/2...ysis-tutorial/.

The method is available on Genestack.

--Misha
P.S. I'm the founder of Genestack, happy to answer questions etc.

[kobeho24]

Originally Posted by ostolop

Hi!

This is a bit off-topic, and somewhat blatant self-advertising, but for what it's worth, we developed a method -- preparing it for publication, draft going on biorxiv soon -- to deal with single cell RNA-seq without ERCC spike-ins.

Here's some info on it http://genestack.com/single-cell-rnaseq/ and a bit more here https://genestack.com/blog/2014/09/2...ysis-tutorial/.

The method is available on Genestack.

--Misha
P.S. I'm the founder of Genestack, happy to answer questions etc.

Hi Misha,
From my point of view, technical noise in single cell experiment such as PCR bias should always be taken into account. As far as I know, ERCC and UMI are the only two options we have to defeat the technical noise to some degree. And I am not sure that the issue can be solved in a bioinformatic manner. Anyway, thank you for your reply. I will take a look at it and see if there is any excellent information for future reference.

Best,
Gary

[ostolop]

Hi Gary,

You're absolutely right -- if you *have* spike-ins, it makes your experiment design more meaningful. We implemented methods that make use of spike-ins, of course, that goes without saying.

We ended up designing and implementing a non-parametric noise-reduction method because we got tasked with processing a dataset that didn't have spike-ins, ERCC or UMI...

We then tested our method on the Brennecke dataset and got high concordance! So, we know it works well..

--Misha

[Nighthawkrao77]

Ostolop - I'd be curious to see what the pros/cons of your platform vs. Seven Bridges vs. DNANexus, etc.... Seems like more and more groups are moving to single-cell analysis with the use of monocle and other public apps.

[Conradona]

Dear all,

I am preparing single-cell libraries according to the method described in Macaulay I, et al. G&T-seq: Parallel sequencing of single-cell genomes and transcriptomes. Nature Methods, 2015.

The amplified cDNA seems to look fine on the tapestation when I add no ERCC spike-in controls.

But I get strange double peaks when I add the controls and the tapestation screentape gets overloaded. The paper says to use 1 μl of a 1:250,000 dilution and I went up to 1:500,000 and 1:1,000,000 but get the same results.

The project leader tells me to just dilute more and more but I am suspicous that this won’t solve the problem.

Thanks in advance for any hints!
Conny

[szhaoaf]

Originally Posted by Conradona

Dear all,

I am preparing single-cell libraries according to the method described in Macaulay I, et al. G&T-seq: Parallel sequencing of single-cell genomes and transcriptomes. Nature Methods, 2015.

The amplified cDNA seems to look fine on the tapestation when I add no ERCC spike-in controls.

But I get strange double peaks when I add the controls and the tapestation screentape gets overloaded. The paper says to use 1 μl of a 1:250,000 dilution and I went up to 1:500,000 and 1:1,000,000 but get the same results.

The project leader tells me to just dilute more and more but I am suspicous that this won’t solve the problem.

Thanks in advance for any hints!
Conny

Hi Conradona,

Recently, I am working with single cell pick with Smart-seq2 method and encounter the same problem as you. Without adding ERCC, the fragment analyzer shows good quality of amplified cDNA. However, after adding ERCC into lysis buffer, a strange double peaks appears as you said, even the same as negative control.

Could you tell how you solve this problem in the end?

Thank you very much.

Best regards.
Alex

[TimMercer]

Dear all,

Also a bit of self-advertising - but we have developed a new set of RNA spike in controls (we call them sequins) that can be used in single-cell RNA sequencing instead of ERCCs.

Unlike ERCCs, they are modeled on human genes, so they include spliced alternative isoforms. Additionally, they have been prepared with care to provide a tighter reference ladder, wider dynamic range, and no-thaw manufacture (to ensure full length transcripts).

They are free for non-profit research, and we can send as much as you need for your experiments. Order them from at www.sequin.xyz/order

We also describe them in detail in the following Nature Methods paper: https://www.nature.com/articles/nmeth.3958

Hope they help!
Tim.