nf-co.re/airrflow v5.0.0 is out! alert Please check out the CHANGELOG for a quick list of changes in this version: https://github.com/nf-core/airrflow/releases/tag/5.0.0

Important additions to this version https://nf-co.re/airrflow/5.0.0/: • Uses pRESTO 0.7.6, which goes back to a less stringent quality filtering of raw reads. For details check the pRESTO log. • Shout out to @Huimin Lyu who improved the single-cell and bulk tutorials, including a video recording of the single-cell tutorial so you can follow it step-by-step, as well as multiple bugfixes and improvements 🎉 • We are now using the nf-test testing framework 🧪

Hi guys! I noticed that the inferred distance threshold used to define clonal relationships differs between my samples. Would these sample-specific thresholds affect the validity of comparing Shannon diversity indices between samples, even though the same clonotype-definition parameters were applied to all of them? Or is this comparison still appropriate under these conditions? Thanks in advance!

Hi all, Some more questions about pipeline performance improvements: 1. During the UMI Build Consensus step, I notice that core usage is low (average load of 2 cores) despite setting high-core usage for this step using config files and the parameter being forwarded to the python job (--nproc 64). At the start of this step, I see that all requested cores are used, but after a brief interval only 1/2 cores remain in use. Why is that? How can this be improved? Runs of large datasets are failing due to reaching time limits, if this step could be sped up it would make a big difference for me. 2. The "argument is of length zero" error of the Collapse Duplicates step is still causing failures in v. 5.0.0 for some of my samples. The workaround I used in version 4.3.0 does not work anymore so I am stuck. Any idea when this issue might be solved? Cheers, Thomas-Wolf

🎉 <!channel> If you are working on AIRR sequencing join us for the AIRR-C meeting this year! Registration is now officially OPEN for AIRR Community Meeting VIII: Decoding and Recoding Immunity! 🙌 📅 June 8–11, 2026 📍 Yale University (US). Virtual option available. 🔗 Register here: https://tinyurl.com/airrcmeeting8 We’ve got an awesome program shaping up — keynotes, posters, deep‑dive tutorials, mentoring sessions, and of course our group dinner on June 10! 🍽️ 💥 Early‑bird pricing is available until April 15, so be sure to grab your spot before then — and don’t forget to use your discount codes if you are an AIRR-C member (you'll get them by email when signing up for the membership). A huge thank you to our early sponsors Cellecta (Gold), Enpicom (Silver), Takara Bio (Bronze), Waters Biosciences formerly BD Biosciences (Bronze), and others we will announce soon. We couldn’t do this without you!

Hi, I have rat data (bulk RNA-seq) -> I cannot use airrflow, correct? thx!

Hi all, I have a question about the pipeline. Is the qc-filtering step particularly slow? It takes me a lot of time every time I run it, but it doesn't seem to take up a lot of computing resources.

Hi all, we’re running the nf-core/airrflow pipeline on a secure offline system. We installed the pipeline in offline mode and made sure all container images are available locally. However, the pipeline still attempts to download external resources, for example: https://raw.githubusercontent.com/nf-core/test-datasets/airrflow/database-cache/igblast_base.zip Since outbound network access is blocked, this obviously fails. My question: Where should these files be placed locally so that the pipeline can pick them up instead of trying to download them? Is there a recommended cache or directory structure for this? Thanks a lot for any guidance!

Hi all, I recently used the Airrflow pipeline to do a fairly involved analysis for a collaborator (with many additional downstream steps). As we were typing up the work, I noticed that I had used the defaults for

--reference_fasta

and

--reference_igblast

. I guess that if you don't also set the

--fetch_imgt

flag, the references are downloaded from here and here instead of using the latest IMGT reference data at runtime. Redoing the analysis isn't really feasible, so I'm trying to identify which release of the IMGT database is included with the defaults and have a few questions and suggestions: 1. The imgtdb_base.zip file contains a

IMGT.yaml

file with this information:

source:  <https://www.imgt.org/genedb>
date:    2024.11.26
species:
    - human:Homo+sapiens
    - mouse:Mus

And the igblast_base.zip file contains

/database/*.njs

files with information such as:

{
  "version": "1.2",
  "dbname": "imgt_mouse_tr_d",
  "dbtype": "Nucleotide",
  "db-version": 5,
  "description": "/home/gisela/Projects/Pipelines/test-airrflow/work/19/d57207f9cc332598ed7c76c66dc189/igblast_base/fasta/imgt_mouse_tr_d.fasta",
  "number-of-letters": 60,
  "number-of-sequences": 5,
  "last-updated": "2024-11-26T14:48:00",
  "number-of-volumes": 1,
  "bytes-total": 49924,
  "bytes-to-cache": 300,
  "files": [
    "imgt_mouse_tr_d.ndb",
    "imgt_mouse_tr_d.nhr",
    "imgt_mouse_tr_d.nin",
    "imgt_mouse_tr_d.nog",
    "imgt_mouse_tr_d.nos",
    "imgt_mouse_tr_d.not",
    "imgt_mouse_tr_d.nsq",
    "imgt_mouse_tr_d.ntf",
    "imgt_mouse_tr_d.nto"
  ]
}

As I understand it, none of these versions (e.g.

db-version: 5

) correspond to the IMGT database. Given the date (2024.11.26), should I refer to this page to infer the state of the database at the time these files were prepared? 2. I'm not finding any information about how these files were prepared (e.g., commands used, etc) in the Airrflow documentation, Airrflow report, or Nextflow logs. To ensure reproducibility, I'd recommend making this information more explicit / readily available. 3. Finally, it would seemingly make sense to have

--fetch_imgt

set to True by default. That would ensure that new users, such as myself, use the optimal files by default.

Hi there!:) I would love to use airrflow, since a couple of days I'm trying to just run:

nextflow run nf-core/airrflow -r 5.0.0  -profile test,docker

but got:

N E X T F L O W   ~  version 26.04.0

WARN: Cannot read project manifest -- Cause:  Config parsing failed
Error nextflow.config:88:25: `pipelines_testdata_base_path` is not defined
│  88 |     reference_igblast = pipelines_testdata_base_path + 'database-cache

I tried also nf-core/rnaseq and it works. I'm using WSL, did someone encountered the same issue or knows how to solve it? Thank you :)

Hi there, would love some assistance with a rhesus macaque custom primer error. I spiked a mix of custom primers into a human 10x vdj library prep kit, and the reverse complement sequences are clearly found in my vdj reference (Mmul10), which I built using cellranger mkvdjref. However, my pipeline is breaking, as below:

Command executed:

  cellranger \
      vdj \
      --id='1486_MLN_BCR' \
      --fastqs=. \
      --reference=Mmul_10_IMGT_vdj \
      --localcores=8 \
      --localmem=16 \
      --inner-enrichment-primers /nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis/nhp_bcr_primers_RC.fasta
  
  cat <<-END_VERSIONS > versions.yml
  "NFCORE_AIRRFLOW:AIRRFLOW:SC_RAW_INPUT:CELLRANGER_VDJ":
      cellranger: $(echo $( cellranger --version 2>&1) | sed 's/^.*[^0-9]\([0-9]*\.[0-9]*\.[0-9]*\).*$/\1/' )
  END_VERSIONS

Command exit status:
  1

Command output:
  Martian Runtime - v4.0.12
  Serving UI at <http://a1130u05n01.mghpcc.ycrc.yale.edu:35431?auth=LbRzWFVVD11hMRrDkaHC1qp9CefwOD7dxm21n6FPInM>
  
  Running preflight checks (please wait)...
  Checking sample info...
  Checking FASTQ folder...
  Checking reference...
  Checking reference_path (/nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis/Mmul_10_IMGT_vdj) on a1130u05n01.mghpcc.ycrc.yale.edu...
  Checking inner enrichment primers...
  Checking enrichment primers (/nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis/nhp_bcr_primers_RC.fasta) on a1130u05n01.mghpcc.ycrc.yale.edu...
  
  [error] None of the C-REGIONs in the reference /nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis/Mmul_10_IMGT_vdj is targeted by the following inner enrichment primer(s): CACTCAACGTGACCTGGAGCAA, CAAAGTCCGTGACTGTCACCTG, AAGCACTATGACCTTACCAGCC, TCCTACAGTCCTCAGGGCT, CCTCGTCTCCTGTGAGAATGC, CTCTGTTGTGTGCCTGCTGAAT, CAAGGCCACACTAGTGTGTCT.

nf-core/airrflow 4.3.0
------------------------------------------------------
Input output options
  input                     : Nine_runs.tsv
  outdir                    : 515_nine_runs_output

Protocol
  library_generation_method : sc_10x_genomics

Single cell analysis options
  reference_10x             : Mmul_10_IMGT_vdj

Clonal analysis options
  clonal_threshold          : 0.2

Institutional config options
  config_profile_name       : mccleary
  config_profile_description: McCleary Cluster at Yale
  config_profile_contact    : Gisela Gabernet, @ggabernet
  config_profile_url        : <https://docs.ycrc.yale.edu/clusters/mccleary/>

Generic options
  trace_report_suffix       : 2026-05-15_15-30-32

Core Nextflow options
  revision                  : 4.3.0
  runName                   : wise_ritchie
  containerEngine           : singularity
  launchDir                 : /nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis
  workDir                   : /nfs/roberts/project/pi_lk595/tar37/NHP_GI/BCR_analysis/work
  projectDir                : /home/tar37/.nextflow/assets/nf-core/airrflow
  userName                  : tar37
  profile                   : mccleary
  configFiles               :

For single cell samples that are inherently sparse in B cells, is there a way to bypass the 5% cutoff seen below?

Log message:
  An extremely low rate of correct barcodes was observed for all the candidate chemistry choices for the input: Sample 1509_ILEUM_BCR_RMT in "/tmp/nxf.DxsNzi5CBT". Please check your input data.
  - 3.8% for chemistry SCVDJ-R2-v3

Hi everyone, I am new to analyzing AIRR-seq data and to nf-core/airrflow, so I hope this is the right place to ask for help. I am trying to analyze data generated with the Takara Bio MaxiSeq Human FFPE TCR+BCR (with UMIs) kit (https://www.takarabio.com/products/next-generation-sequencing/immune-profiling/human-repertoire/maxiseq-human-ffpe-tcrbcr-(with-umis)) from FFPE tumor samples So my questions are: 1. Does anyone have experience running airrflow with this specific Takara Bio MaxiSeq FFPE kit, and could you share what primer FASTA files / parameters you used? 2. Regarding the primer FASTA files themselves: the kit does not provide the C-region primer sequences explicitly in the user manual, and there are no V-region primers in the PCR design (the kit uses a universal extension primer instead). How would you recommend obtaining or constructing the

--cprimers

and

--vprimers

files needed for the analysis? Is there a public resource or a standard workaround for commercially closed primer pools like this one? Any guidance would be greatly appreciated. Thanks in advance!

Hello, I am running an analysis and the FIND_CLONAL_THRESHOLD is taking forever, this is what I have on my nextflow.config: withName: 'FIND_CLONAL_THRESHOLD' { cpus = 1 memory = { 128.GB * task.attempt } time = { 48.h * task.attempt } queue = 'week' errorStrategy = { task.exitStatus in [137, 140] ? 'retry' : 'finish' } maxRetries = 3 } Is there any way to accelerate this or should I just increase the time wait it out? Thanks,

Hi Ramiro @Ramiro Barrantes Reynolds, although FIND_CLONAL_THRESHOLD has intensive distance calculation, but it never took noticeably long time for my analysis. In your configuration, you request 128GB * task.attempt from on cpu. Are you sure you have a single cpu with such high amount of memory available? Probably the real memory is much lower than the one you request. Try to allocate more CPUs to take advantage of parallel processing and allocate reasonable memory based on your system. For your reference, the default configuration for find_clonal_threshold is:

time = { 30.h * task.attempt } cpus = { 16 * task.attempt } memory = { 72.GB * task.attempt }

Hi Huimin @Huimin Lyu, thank you, it seems to me that FIND_CLONAL_THRESHOLD is serial. We do have a single CPU with a lot of memory. I gave it more CPUs but the process does not seem to be using them, and I don't see any configuration about that online

✅ airrflow/test_full completed successfully! 🎉 Run Name mighty_hugle_2 Duration 2h 46m 11s Status ✅ Success Tasks Completed: 274 Jun 8, 2026 at 11:30 AM

nf-co.re/airrflow v5.1.0 is out!!!

Please check out the CHANGELOG for a quick list of changes in this version: https://github.com/nf-core/airrflow/releases/tag/5.1.0

Hey there

Will anyone here be in Prague next week for the yEFIS Symposium?

Hi everyone, @Gisela Gabernet @Huimin Lyu I have two PRs open on nf-core/airrflow and would be grateful for a review whenever you have time: github.com/nf-core/airrflow/pull/506 : splits the SingleCellQC step into HeavyChainFilter and RemoveContaminants github.com/nf-core/airrflow/pull/496 : adds a sequencing depth plot to the airrflow report PR #497 has already been approved, but I can not merge it into dev. I would appreciate any advice on how to resolve this. Thank you in advance!

Hi all. Thanks for developing this nice pipeline. I have written a bug report in the GitHub repo (github.com/nf-core/airrflow/issues/507) and would appreciate if someone could check if what I'm observing is indeed the case.

Hi, my Claude session seems to think that if I have multiple rows of my samplesheet (ie multiple lanes) corresponding to the same sample, and want to run via TRUST4, that it will not concatenate my multiple FASTQ pairs and instead only take the first one (subworkflows/local/fastq_input_check.nf, lines 33–36). Can someone please confirm and if so, is it possible to incorporate a fix? In my experiment I have multiple lanes of a standard 5' 10x (unenriched) GEX library per sample and would want to have those merge properly

The comment explains the skip for cellranger, but trust4 is included as well. Rows with the same sample_id are grouped, but CAT_FASTQ never runs. The sample moves on as one list: [R1_L1, R2_L1, R1_L2, R2_L2, …].

modules/nf-core/fastp/main.nf, which runs next: only the first two files are used.

ln -sf ${reads[0]} ${prefix}_1.fastq.gz
ln -sf ${reads[1]} ${prefix}_2.fastq.gz

That's lane 1’s R1 and R2. Even if all lanes reached TRUST4, modules/nf-core/trust4/main.nf (lines 42–43) passes only forward[0] and reverse[0].

The result is no error, just a TRUST4 run on lane 1 alone. With 4 lanes, you'd lose about three quarters of the reads.

we are working on upgrading the megatest submission. I am using airrflow as a testing ground because it uses the v2 syntax parser. Feel free to ignore the PRs at the moment.

❌ airrflow/test_full failed 😿 Run Name marvelous_heyrovsky Duration 25s Status ❌ Failed Error Message

Missing process or function paramsSummaryLog([[:], repository: <https://github.com/nf-core/airrflow>, projectDir: /.nextflow/assets/.repos/nf-core/airrflow/clones/a67dad75f6a1cc31aa780b4804760fbb44833912, commitId: a67dad75f6a1cc31aa780b4804760fbb44833912, revision: null, startTime: 2026-10-15T09:50:44.833482479Z, endTime: null, duration: null, container: {}, commandLine: nextflow run '<https://github.com/nf-core/airrflow>' -name marvelous_heyrovsky -params-file '<https://api.cloud.seqera.io/ephemeral/--59dOBXKzqVAxkQqtOZ2g.json>' -with-tower -r a67dad75f6a1cc31aa780b4804760fbb44833912 -profile test_full, nextflow: nextflow.NextflowMeta(version:26.04.6, build:12646, timestamp:09-07-2026 18:49 UTC, preview:nextflow.NextflowMeta$Preview@5a83be9, enable:nextflow.NextflowMeta$Features@6a9781fd, strictModeEnabled:false, moduleBinariesEnabled:false), success: false, workDir: /nf-core-awsmegatests/work/airrflow/work-a67dad75f6a1cc31aa780b4804760fbb44833912, launchDir: /, profile: test_full]) -- Did you mean 'paramsSummaryLog' instead?

Command Line

nextflow run '<https://github.com/nf-core/airrflow>' -name marvelous_heyrovsky -params-file '<https://api.cloud.seqera.io/ephemeral/--59dOBXKzqVAxkQqtOZ2g.json>' -with-tower -r a67dad75f6a1cc31aa780b4804760fbb44833912 -profile test_full

Oct 15, 2026 at 9:51 AM

✅ airrflow/test_full completed successfully! 🎉 Run Name distraught_moriondo Duration 2h 42m 47s Status ✅ Success Tasks Completed: 274 Oct 15, 2026 at 12:41 PM

depictio Depictio dashboard for airrflow: feedback wanted Depictio (open source, from EMBL) turns the output of a workflow run into an interactive dashboard you can share, so nobody has to write a Shiny or Streamlit app for each project. Point a template at any run of a pipeline and you get the same dashboard, which you can edit (UI & YAML), version and comment. 🧬 The airrflow dashboard template (no account needed) shows V gene usage, CDR3 spectratype, clonal diversity and expansion, and clones shared between a subject's samples. 🙌 We need your help! Feedback and ideas welcome in #C0AFS06G9HR (all templates and details there) or in the airrflow GitHub discussion.

Hi all! Does anyone have a strategy for finding public clonotypes across subjects? We have done scBCR-seq from 3 subjects with two visits each (pre- and post-vaccination), 6 samples total, run through airrflow 5.0.0. in

--mode assembled

from CellRanger output. Clonal inference was done per subject (both visits together) with automatic threshold detection. What I’m unsure about is the public clonotype analysis step: 1. Is this something the pipeline can do? How should the samplesheet be organised for it — all 6 samples with subject IDs as they are, or does the grouping need to change so clones get inferred across subjects? 2. Is re-running clonal inference on all subjects pooled the right move, or does that break the threshold inference? Each subject currently has its own automatically detected threshold, so I’m not sure whether to pool and re-infer, or to compare the per-subject clonotypes in a separate step under a single fixed criterion. 3. If this is better done outside the pipeline, any recommended route? Thanks in advance!