--assemble-contigs-by what to specify to run mixcr findAlleles

[analeigh-gui]

I'm following the guide: Immunoglobulin lineage trees reconstruction

I have BD Rhapsody full-length BCR mouse data. I first run mixcr analyze using the preset.

mixcr analyze bd-rhapsody-mouse-tcr-full-length \
        20220923-Mm023B-BCR_R1.fastq.gz \
        20220923-Mm023B-BCR_R2.fastq.gz \
       mouse_vdj/result

Then I ran:

mixcr findAlleles \
        --report findAlleles.report.txt \
        --json-report findAlleles.report.json \
        --export-alleles-mutations alleles.tsv \
        --export-library alleles.json \
        --output-template reassigned.clns \
        result.contigs.cls

And got the message:

Input files must not be processed by assembleContigs without --assemble-contigs-by option 

It is unclear from the manual what options the --assemble-contigs-by can take. Could you recommend it? Thanks!

[PoslavskySV]

Hi,

BD full-length protocol is fragmented, so MiXCR uses contig assembly to reconstruct the longest-possible contig sequence (VDJRegion plus L1,L2+UTR in V part in the best case). As it is not know a priori which region will be fully covered for each cell, the length of the resulting clonotype will be varying, which is not acceptable for allelic variant inference.

That said, you can do the following:

mixcr analyze bd-rhapsody-mouse-tcr-full-length \
        --assemble-contigs-by VDJRegion \
        20220923-Mm023B-BCR_R1.fastq.gz \
        20220923-Mm023B-BCR_R2.fastq.gz \
       mouse_vdj/result

Here the option --assemble-contigs-by VDJRegion will drop all clones which do not cover VDJRegion and for those which cover a longer feature it will cut them to just VDJRegion.

If you don't want to re-run full analyze command, you can re-run only assembleContigs step on .clna input files:

mixcr assembleContigs --assemble-contigs-by VDJRegion input.clna output.clns
mixcr exportClones output.clns output.tsv

(in next releases we will improve analyze so that it will not re-run all underlying steps)