Question about long read paramter

[e-antoun]

Hi,

I'm running MiXCR on some long read ONT data with cell barcodes and UMIs, using a custom preset (attached).

ont-modified-v1.yaml

I was just wondering about one the parameters and what exactly it is used for, as I couldn't find any information about it. In the align step, there is:

    longReadsParameters:
      windowSize: 1500

It seems if I change this value, it does affect my final results so just hoping to get an idea of what it is doing. The preset I'm using is one from #1681

Thanks

[mizraelson]

Hi, this is a parameter of aligner that splits long reads into chunks, within each "window" a k-mer mapping search is performed. What data do you have and how changing the parameter affects the output?

[e-antoun]

Hi, this is a parameter of aligner that splits long reads into chunks, within each "window" a k-mer mapping search is performed. What data do you have and how changing the parameter affects the output?

I have some ONT sequencing of TCR enriched libraries generated from the 10X Visium workflow. There are reads in the FASTQ file that align to a certain clonotype that I know is in the sample, when I manually IgBLAST those reads, they align perfectly to the V-J genes with a few expected differences in the CDR3 sequence. We also know this is a real clonotype as we have paired scRNAseq data for this sample with VDJ sequencing so we know >5K clonotypes from the same tissue (adjacent tissue section) and the same patient

However, running the preset with windowSize: 1500, this clonotype doesn't appear, and instead, the reads that align to this clonotype get assigned to a nonsensical clonotype, but when I increase the windowSize to 2500, that and several other clonotypes now do appear in the final clonotype table:

windowSize: 1500 -

cat clones_TCR_alignments.tsv | grep "da413eb8-58e3-4130-a4e7-5b053295bf81" 

TRAV12-2                TRAJ23  TRAC    SWWHCRGGTEVV    CTGGGCCTGCTATATC+CACAGGGTCCTC: 1.0      TRA     true    -1:16   -1      1010595 1:1616|da413eb8-58e3-4130-a4e7-5b053295bf81 qs:f:26.0681st:Z:2025-01-21T14:15:16.453+00:00      RG:Z:a609bf2a1419d4c0dbb78fb24b5b0a682f129301_dna_r10.4.1_e8.2_400bps_sup@v5.0.0        DS:Z:gpu:Quadro RTX 8000 strand=+

windowSize: 2500 -

zcat "runsCombined_MiXCR_TCR_alignments.tsv.gz" | grep "da413eb8-58e3-4130-a4e7-5b053295bf81"

TRAV12-2                TRAJ23  TRAC    CAAYNQGGKLIF    CTGGGCCTGCTATATC+CACAGGGTCCTC: 1.0      TRA     true    -1:16   -1      1010595 1:1616|da413eb8-58e3-4130-a4e7-5b053295bf81 qs:f:26.0681st:Z:2025-01-21T14:15:16.453+00:00      RG:Z:a609bf2a1419d4c0dbb78fb24b5b0a682f129301_dna_r10.4.1_e8.2_400bps_sup@v5.0.0        DS:Z:gpu:Quadro RTX 8000 strand=+

Exact same read, two completely different CDR3 clonotypes, with the bottom one the correct one when mapping it back to the matched single-cell VDJ data and manually annotating it using IgBLAST

Thanks

[mizraelson]

Hm, can you please share the original read da413eb8-58e3-4130-a4e7-5b053295bf81

[e-antoun]

Hm, can you please share the original read da413eb8-58e3-4130-a4e7-5b053295bf81

Please see below for the read:

@1:1616|da413eb8-58e3-4130-a4e7-5b053295bf81 qs:f:26.0681       st:Z:2025-01-21T14:15:16.453+00:00      RG:Z:a609bf2a1419d4c0dbb78fb24b5b0a682f129301_dna_r10.4.1_e8.2_400bps_sup@v5.0.0       DS:Z:gpu:Quadro RTX 8000 strand=+
CTACACGACGCTCTTCCGATCTCTGGGCCTGCTATATCCACAGGGTCCTCTTTTTTTTTTTTTTTTTTTTTTTTTTTTGTTCTTCCAAATCATTTAAATGAAGGCATCGGCAGCAGGACATACAACACGCCCAGACACACAGCATAGCAGGGCCTCGATAATGAGATAATTTCCCCCCACGTCTTGAGAAGAAGAATACTATGTATTTCTTTATGAACACTATTAAAAAAAAATAAACCCCTCACAACATTCTCCAGGACCTAGAGCCCAAGAGAACCCACTGAAGATCCATCATCTGTGGGATGGCGGAGGCAGTCTCTGGGGAGCAGGAGGGAATGTGCACAGCCAGGGGAGGCTGCAGCAGCCTTGCCTCTGCCGTGAATGTCAGGCAGTGACAAGCAGCAATAAGGGAACAGAGGGGGTGGCAGCAGTGTTTGGCAGCTCTTCAGCAATCTTAATCATAAATTCGGGTAGGATCCAGTTGGTGGCATTGCCGGGGGGGCACAGAGGTGGTAGCAGCTTTCACCTCCTTGGGGGTGGGAGAGTTCCCTCTGTTTGGAGAGGGAGAAGAGGGGTAATGCAGAGGAAGGAGCGAGGGAGCACAGGCTGTCTTACAATCTTGCAGATCTCAGCTGGACCACAGCCGCAGCGTCATGAGCAGATTAAACCCGGCCACTTTCAGGAGGAGGATTCGGAACCCAATCACTGACAGGTTTTGAAAGTTTAGGTTCGTATCTGTTTCAAAGCTTTTGTCCGACCAGCTTGACATCACAGGAACTTTCTGGGCTGGGGAAGAAGGTGTCTTCTGGAATAATGCTGTTGTTGAAGGCGTTTGCACATGCAAAGTCAGATTTGTTGCTCCAGGCCACAGCACTGTTGCTCTTGAAGTCCATAGACCTCATGTCTAGCACAGTTTTGTCTGTGATATACACATCAGAATCCTTACTTTGTGACACATTTGTTTGAGAATCAAAATCGGTGAATAGGCAGACAGACTTGTCACTGGATTTAGAGTCTCTCAGCTGGTACACGGCAGGGTCAGGGTTCTGGATATTGGGTTTCACAGATAACTCCGTTCCCTGTCCGAAGATAAGCTTTCCTCCCTGGTTATAAGCGGCACAGAGGTAGGTGGCTGAATCACTGGGCTGGGAGTCTCTGATGAGCAGAGAAACATACTGGCTGGCTTTATTGAGCTGTGCTGTAAACCTTCCATCTTCTTTGTCACCATTGGAGTATATGAACATTATCAACTCAGGGCTTTTCCCAGAATATTGTCTGTACCAGAAGAAGGACTGGGAACCTCGGTCACTGTAAGTGCAGTTGAGAGAGGCAATGGCTCCCTCTGGAACACTGAGGGGTCCAGAATTCTGCTCCACCTCCTTCTGTTGGCTCCAAACCCAGCTCAACTGAAGCCACAGGATCACTAGTAAAACTCTCAAGGATTTCATCATTCTTCTCTGCCCTGGTTCACTGAGGATTCAAATTTTAAACCCTTAAAAAATGCACCTAGAATGGATATTTCTACTTTAAAAATCATAAAAAGAATCTGCTTTACCAGTTCAAAAAAATGATGCTCTTTTTAACCCCCATGTACTCTGCGTTGATACCACTGCTT

Thanks for looking into this