Using MiXCR single-cell preset without cell tags

[imbertj]

Hello,
I'm trying to run the following script using the generic low-throughput single-cell fragmented preset

# directories
input_dir=out/minibams/PD63118/make_fastq
output_dir=out/minibams_rmdup_mixcr/sc_dna
mkdir -p $output_dir

# mixcr
/usr/lib/jvm/java-11-openjdk-amd64/bin/java -Xmx5g -Xms3g -jar /nfs/users/nfs_j/ji4/mixcr/mixcr.jar analyze generic-lt-single-cell-fragmented \ 
  --species hsa \ 
  --assemble-longest-contigs \
    $input_dir/49882_plex2.IGandTR.r1.fastq \
    $input_dir/49882_plex2.IGandTR.r2.fastq \
    $output_dir/result

I'm getting this error message:

Must contain at least one Cell tag (determined as tag name starting from "cell" (like "CELL1", "Cell", "CellId", etc..)).

I don't have cell tags but I have a pair of fastq files for each cell. How can I edit my script to reflect this? The MiXCR website suggests using {CELL:a} but I don't understand what this means.

Second error message:

MiXCR requires --assemble-contigs-by or --assemble-contigs-by-Cells
Why isn't --assemble-longest-contigs working for this preset?

Thanks!

[mizraelson]

Hi,
This preset requires a CELL tag, which in your case can be extracted from the filename. For example, if you have a pair of FASTQ files for each well:

sample_A1_R1.fastq.gz
sample_A1_R2.fastq.gz
sample_B1_R1.fastq.gz
sample_B1_R2.fastq.gz
sample_C1_R1.fastq.gz
sample_C1_R2.fastq.gz
sample_D1_R1.fastq.gz
sample_D1_R2.fastq.gz

You can specify the input as:

mixcr.jar analyze generic-lt-single-cell-fragmented \ 
  --species hsa \ 
  --assemble-contigs-by-cells \
    sample_{{CELL:a}}_{{R}}.fastq.gz \
    $output_dir/result

In this case, the part of the filename that varies between wells (e.g., A1, B1, C1, D1) will be used as the cell tag. The tag can be any portion of the filename that uniquely identifies a well.

The final step of the pipeline, mixcr assembleCells, groups cells into sets based on the clonal sequences they contain. This step requires all clones to cover the same feature; otherwise, we cannot determine whether to aggregate, for example, cell A, which covers FR1 to FR4, with a group of cells that have identical sequences from FR2 to FR4 but are missing FR1 and CDR1.

To address this, you can either:
• Use --assemble-contigs-by-cells, which disables the mixcr assembleCells step and assembles the longest possible sequence for each cell.
• Use --assemble-contigs-by <Feature> to assemble all clones by the same feature and group the cells accordingly.