Recommended Preset for Single-Cell B Cell Receptor (Ig) Data from SMART-Seq v4?

[gsr9999]

Hello MiXCR Team and @mizraelson,

Thank you for developing and maintaining such a wonderful and valuable tool/mixcr. I greatly appreciate this.

I have a question about the appropriate MiXCR preset for our single-cell BCR Ig data (B Cell Receptor).

We sort one human B cell per well in a 96 well plate, generate cDNA using the Takara SMART-Seq v4 Ultra Low Input RNA Kit (oligo-dT + template switching, no UMI), followed by PCR-based Ig enrichment and Illumina sequencing. Each well has its own FASTQ pair, with no embedded cell barcode.

I tested 3 wells using both rna-seq and smart-seq2-vdj (MiXCR v4.7.0-370-develop). Both identify the same dominant clonotypes, but rna-seq produces many low-read-count error-like clonotypes, whereas smart-seq2-vdj collapses these and gives results much closer to the expected 1 productive IGH + 1 productive IGK/IGL per cell. smart-seq2-vdj also gives a substantially higher alignment rate and higher read count.

Since our protocol is SMART-Seq v4, not SMART-Seq2, is smart-seq2-vdj still the recommended preset for this type of single-cell data? Or would you recommend rna-seq with specific parameter adjustments?

Thank you,
srg

[mizraelson]

Hi, in this case the best command would be:

mixcr analyze generic-lt-single-cell-amplicon \
    --species hsa \
    --rna \
    --floating-left-alignment-boundary \
    --floating-right-alignment-boundary C \
    --assemble-clonotypes-by VDJRegion \
      sample_{{CELL:a}}_{{R}}.fastq.gz \
      result

the {{CELL:a}} tag marks the part of the filename that describes the well, for example:

sample_A01_R1.fast.gz
sample_A01_R2.fast.gz
sample_A02_R1.fast.gz
sample_A02_R2.fast.gz
sample_A03_R1.fast.gz
sample_A03_R2.fast.gz
...
sample_H12_R1.fast.gz
sample_H12_R2.fast.gz

VDJRegion implies that the full receptor sequence is covered