10x genomics and 'paired end' reads

[zauhar]

Hi all,

We recently started looking at single cell RNA-seq data produced by our sequencing facility. There were some questions about alignment quality, so I was looking at the fastq read data. The data was generated by 10x genomics library prep (not sure of the specific microfluidic hardware) followed by sequencing on an Illumina machine.

The project PI requested a 'paired end' experiment, and imagined we would have data similar to that provided by a MiSeq experiment, with forward and reverse reads of same length.

Instead, the R2 file has full length reads, and the R1 goes only far enough to capture the cell barcode and UMI (26 bases).

The protocol followed only used 26 cycles for R1. When I questioned why the R1 reads are 'truncated' to minimal length, the manager of the facility basically answers 'that's the 10x protocol'.

We are new to single cell and the 10X platform - is there any reason why more sequence information is not collected on the 'R1' side?

Thanks!
Randy

[zauhar]

This may answer my own question - the facility manager sent a followup: the 10X library prep involves a 30nt poly-T section after the barcode/UMI, and he indicates that the 10X analysis software will ignore anything past the barcode and UMI.

So I guess the implicit reasoning is that it's not worth so many cycles to get to fragment sequence. Does that correctly sum it up?

Randy

[GenoMax]

Did your facility use software from 10x genomics (cellranger mkfastq) to do the demultiplexing of the data? You should ideally use the entire pipeline to process the data from sequences to counts at end. You can find information about the pipeline 10x supplies here.

[zauhar]

Yes, they did do the whole analysis, including generation of alignments, counts and clustering.

I have also installed cellranger on my own server so I can better understand their pipeline.

[GenoMax]

I have not done single cell data for 10x but it sounds like your core folks know what they are doing.

[HESmith]

Read 1 begins with the RT primer. It contains the cell barcode, UMI, and oligo(dT), which primes 1st strand synthesis from the mRNA's poly(A) tail. Until you get past the poly(A), which can be long, there's no useful information to be gained from sequencing past the UMI.

[arkanion]

the manager of the facility basically answers 'that's the 10x protocol'

that's the way a manager sounds like

by the way, you can do 10x single cell rna-seq analysis by yourself here: CSI NGS Portal

[SeqCharlotte]

We recently ran 10x libraries as 150bp paired end, and safe to say that after about 28 base pairs on the first read, the data dropped off completely. We panicked at first but then confirmed with Illumina and 10X that what we saw was as expected.

[mama]

Q30 read2

How can you avoid getting a low Q30 of read2 (reverse read) in a 150bp paired end run when you sequence 10x sc 3'RNAseq?