Low alignment rates for randomly shredded TCR-B library with MixCR

Question

[pm2012]

Hello community,

Apologies for cross-posting my query here along with the github page for MixCR. I am hoping to seek help from a wider community here.

I recently applied the MiXCR pipeline in 'shotgun' mode to analyze my randomly shredded TCR-B library. However the alignment rates for my full set of samples is lowish (20-37%). As I can see from the report, it seems 'Alignment failed because of absence of CDR3 parts' seems to be the main reason for the low alignment (see part of the report for one of the samples below:

Total sequencing reads: 358155 Successfully aligned reads: 73769 (20.6%) Paired-end alignment conflicts eliminated: 1142 (0.32%) Alignment failed, no hits (not TCR/IG?): 20007 (5.59%) Alignment failed because of absence of CDR3 parts: 257575 (71.92%) Alignment failed because of low total score: 6804 (1.9%) Overlapped: 323 (0.09%) Overlapped and aligned: 174 (0.05%) Alignment-aided overlaps: 106 (60.92%) Overlapped and not aligned: 149 (0.04%) V gene chimeras: 35 (0.01%) J gene chimeras: 34 (0.01%) TRA chains: 1 (0%) TRB chains: 73759 (99.99%) TRG chains: 9 (0.01%)

Here is the command that I am using:

mixcr analyze shotgun -s hsa --starting-material rna --receptor-type trb --contig-assembly --only-productive --align "-OallowPartialAlignments=true" --assemble "-OaddReadsCountOnClustering=true" --export "-m 2 -vGene -nFeature CDR3 -aaFeature CDR3 -dGene -jGene -lengthOf CDR3 -defaultAnchorPoints -count" $baseDir/S59_trimmed_R1.fastq $baseDir/S59_trimmed_R1.fastq S59_trimmed

Would you have any input on how to improve my alignment rate?

Answer 1

[mizraelson]

Hi, The first thing i would suggest is to try our new MiXCR series 4 which has some major updates, e.g.:

mixcr analyze rnaseq-full-length \
    --species hsa \
      input_R1.fastq.gz \
      input_R2.fastq.gz \
      result

More on that here: https://docs.milaboratories.com/mixcr/reference/overview-built-in-presets/#rna-seq-data

What kind of fragmentation did you use? Is it possible that data is indeed enriched with non CDR3 fragments?

I would also add: --not-aligned-R1 notaligned_R1.fastq --not-aligned-R2 notaligned_R2.fastq

parameters to export non aligned reads and manually inspect what part of TCR genes they cover.

In general that sounds like a wet lab issue. Feel free to contact us for support via e-mail: [e-mail]