n00b Smart-seq2 questions

[chilipep]

Hi there, I'm new to Smart-seq and have a couple questions:

1. Can the dT(30)VN primer be substituted with dT(20)VN, dT(23)VN, etc.?
2. How does Superscript IV compare to SSII or SSIII?
3. Is KAPA HiFi HotStart still the best option for preamplification?

Finally a very basic question...
If I am doing targeted amplification of a single gene, can I just use my custom primers for PCR after cDNA preamp and skip all the tagmentation steps? I just want to see whether a heterozygous mouse is picking one allele and silencing the other. I don't need the high resolution of NGS, but instead will TOPO clone these single amplicons and pick enough representative clones to decide whether the alleles are being equally expressed or not.

Thanks for your patience with my n00b self!

[cmbetts]

1. Sure, you can swap in different length RT primers without a noticeable impact on performance
2. SSII is definitely better than SSIII, but I don't know about IV
3. I personally like Takara's SeqAmp, but I used to work there and am biased

A followup question. Why are you adding the cost/complexity of preamplification when gene specific RT-PCR works at the single cell level? Banking the cDNA or assessing multiple genes per cell?

Also, are you sure you don't need NGS level resolution for when you're describing? Single cell methods are going to have lots of allelic dropouts that have nothing to do with actual biological regulation, just good ol' sampling noise, especially for anything that isn't highly expressed. The cost of the number of cells, TOPO cloning, and Sanger reactions required to get a remotely robust answer is going to approach the cost of an NGS run real fast (especially if you use targeted amplicons).

Regardless of sequencing method, I also highly recommend putting UMIs in your RT primer and targeting the 3'UTR, RACE style, if there distinguishing alleles close enough to the end of the transcript. Just watch out for alternative polyA usage and bad annotation. Otherwise, you may falsely assume you've reached a high enough count to assess statistical significance when you've really just been looking at a much smaller population that won the sampling lotto (and PCR jackpot)

[chilipep]

Thanks for all the advice, I really appreciate it.

The reason I'm getting total cDNA from single cells (primary B cells in this case) is for the reason you mentioned: banking the cDNA for analysis of multiple genes.

Okay, thanks for letting me know. I am screening single B cells for their allelic usage. I want to see whether they commit to one allele, or co-express the alleles. I can discriminate between the alleles by a simple RE digest, so I was planning to TOPO clone, and do colony PCR/digest ~4-5 clones per cell, rather than Sanger sequencing. I'm not concerned about the precise ratio, but more interested in whether single cells completely favor one allele or not. Is there an easier way to do this? Basically, I'm most interested in using Smart-Seq2 for its library prep.

What do you mean by "bad annotation"?

1. Sure, you can swap in different length RT primers without a noticeable impact on performance
2. SSII is definitely better than SSIII, but I don't know about IV
3. I personally like Takara's SeqAmp, but I used to work there and am biased

A followup question. Why are you adding the cost/complexity of preamplification when gene specific RT-PCR works at the single cell level? Banking the cDNA or assessing multiple genes per cell?

Also, are you sure you don't need NGS level resolution for when you're describing? Single cell methods are going to have lots of allelic dropouts that have nothing to do with actual biological regulation, just good ol' sampling noise, especially for anything that isn't highly expressed. The cost of the number of cells, TOPO cloning, and Sanger reactions required to get a remotely robust answer is going to approach the cost of an NGS run real fast (especially if you use targeted amplicons).

Regardless of sequencing method, I also highly recommend putting UMIs in your RT primer and targeting the 3'UTR, RACE style, if there distinguishing alleles close enough to the end of the transcript. Just watch out for alternative polyA usage and bad annotation. Otherwise, you may falsely assume you've reached a high enough count to assess statistical significance when you've really just been looking at a much smaller population that won the sampling lotto (and PCR jackpot)

[cmbetts]

By bad annotation, I mean that the polyadenylation sites in RefSeq/Gencode are often incorrect, which can make it difficult to amplify the 3'UTR (Refseq in particular tends to choose the longest 3' UTR EST). Not a problem when you anchor your primers in the CDS, but that's often too long for Illumina. There's a few genome browser tracks for mouse and human specifically for polyA-seq that are a better resource.

I'd be especially concerned about sampling bias amplifying off of primary immune cells. Those cells tend to have very low RNA content, so you'll be lucky just to detect your gene(s) of interest, which likely will represent single captured transcripts. In that case, numbers of cells would be more important than clones per cell. There's almost certainly scRNA-Seq data available for mouse PBMCs that you can use to get a sense of the detection and expression rates you should be expecting and base your cell/colony numbers off of (10X or Drop-Seq 3' sequencing would also give you the correct polyA sites if you decide to try a targeted NGS approach).