many questions about the MALBAC-RNA method

[Simone78]

Dear all,
does anybody have some hands-on experience with the new MALBAC-RNA published some months ago by Sunney Xie´s lab (Chapman et al, PLoS ONE 2015)?

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0120889

If you, among other things, look at Fig 2C they have close to 90% sensitivity (!) between technical replicates at FPKM>1.

My knowledge of single-cell RNA-seq (please correct me if I´m wrong) tells me that most of the transcripts are "lost" in the RT step due to inefficiencies of the reaction. Given that all the methods use a retroviral-based reverse transcriptase they all share this problem and that´s why the sensitivity for lowly-expressed transcripts is never higher than 35-40% (this is the value we get with Smart-seq2, at least). Of course the following PCR also introduces a bias and the fewer cycles one is using the better the results

In this paper they use Superscript III for the RT and only afterwards do they apply 10 cycles of a modified MALBAC reaction. They even do 19 more cycles of PCR after the MALBAC which one might think will increase the bias, rather than making things better.

Any idea/comments anyone? Is it really THAT good? that would be great!
thanks!

[Nighthawkrao77]

Hi Simone - be advised that in least in this paper they use 100 cells in 1 uL and into 4 uL lysis buffer. Having a population of cells could be why their difference between technical replicates is so small.

Also - I have heard (through the grapevine) that there are some issues with bioinformatic analysis that actually expose biases that are not readily apparent. I can't comment on the prep myself having never done it.

[Simone78]

Hi Simone - be advised that in least in this paper they use 100 cells in 1 uL and into 4 uL lysis buffer. Having a population of cells could be why their difference between technical replicates is so small.

Hi,

yes, in the paragraph "Cell culture and sample preparation before single cell amplification" they say: "...a final concentration of 100 cells/uL is reached. 1 uL of the well-mixed diluted cell suspension is added into a total of 4 uL cell lysis buffer".

However, they also have 9 SW480 single cells and in the "Results and discussion" part I read (very beginning): "each cell is picked and transferred into PCR reaction tubes preloaded with mild cell lysis buffer...". As I understand, they pick very few single cells (why not picking more? it takes no time even if you would do it manually...or FACS...or even limited dilution...mystery!) as well as split the cell lysate into single cell "portions".

But, even with technical replicates, having a sensitivity close to 90% for genes with FPKM=1 (Fig 3C) seems "optimistic", to use an euphemism...Am I wrong?

[kobeho24]

Dear all,
does anybody have some hands-on experience with the new MALBAC-RNA published some months ago by Sunney Xie´s lab (Chapman et al, PLoS ONE 2015)?

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0120889

If you, among other things, look at Fig 2C they have close to 90% sensitivity (!) between technical replicates at FPKM>1.

My knowledge of single-cell RNA-seq (please correct me if I´m wrong) tells me that most of the transcripts are "lost" in the RT step due to inefficiencies of the reaction. Given that all the methods use a retroviral-based reverse transcriptase they all share this problem and that´s why the sensitivity for lowly-expressed transcripts is never higher than 35-40% (this is the value we get with Smart-seq2, at least). Of course the following PCR also introduces a bias and the fewer cycles one is using the better the results

In this paper they use Superscript III for the RT and only afterwards do they apply 10 cycles of a modified MALBAC reaction. They even do 19 more cycles of PCR after the MALBAC which one might think will increase the bias, rather than making things better.

Any idea/comments anyone? Is it really THAT good? that would be great!
thanks!

Hi Simone,
If I remember correctly, the pre-amplification in MALBAC is a quasilinear amplification to get a relatively higher amount of DNA with high fidelity. Once the abundance is relatively high enough, there should not be a big issue in terms of PCR bias. Besides, according to my knowledge, the sensitivity of RNA-seq is mainly due to the cDNA synthesis rate(~40% in SMART-seq2 as you said). Therefore, MALBAC-RNA's sensitivity becomes how effective the SSIII RT is in cDNA synthesis with poly-dT primers. MALBAC is just a way that can reduce the PCR bias to some degree, no matter the template is gDNA or cDNA, on my perspective. 90% is so ideal to single-cell RNA-seq that I can not convince myself of it, maybe it's just 100-cell's data instead of single-cell?

Best!
Gary

[kobeho24]

Dear all,
does anybody have some hands-on experience with the new MALBAC-RNA published some months ago by Sunney Xie´s lab (Chapman et al, PLoS ONE 2015)?

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0120889

If you, among other things, look at Fig 2C they have close to 90% sensitivity (!) between technical replicates at FPKM>1.

My knowledge of single-cell RNA-seq (please correct me if I´m wrong) tells me that most of the transcripts are "lost" in the RT step due to inefficiencies of the reaction. Given that all the methods use a retroviral-based reverse transcriptase they all share this problem and that´s why the sensitivity for lowly-expressed transcripts is never higher than 35-40% (this is the value we get with Smart-seq2, at least). Of course the following PCR also introduces a bias and the fewer cycles one is using the better the results

In this paper they use Superscript III for the RT and only afterwards do they apply 10 cycles of a modified MALBAC reaction. They even do 19 more cycles of PCR after the MALBAC which one might think will increase the bias, rather than making things better.

Any idea/comments anyone? Is it really THAT good? that would be great!
thanks!

Hi Simone,
Can the 35-40% transcipt detection sensitivity of SMART-seq2 due to the low efficiency of template switching instead of the low processivity of retroviral RT? If I remember correctly, the cDNA synthesis efficiency of STRT is ~10% while in STRT-C1 has strongly improved to 48%. I suppose performing SMART-seq2 would have the same effect. Am I right?

Best!