MultiQC Dragen abort: division by zero for empty sample

[stefanhelfert]

Description of bug

╭───────────────────────────────────────────── Oops! The 'dragen' MultiQC module broke... ─────────────────────────────────────────────╮
│ Please copy this log and report it at https://github.com/ewels/MultiQC/issues                                                        │
│ Please attach a file that triggers the error. The last file found was: ./z04748_syw1/z04748_syw1.wgs_fine_hist.csv                   │
│                                                                                                                                      │
│ Traceback (most recent call last):                                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/multiqc.py", line 651, in run                                   │
│     output = mod()                                                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/dragen.py", line 64, in __init__                 │
│     samples_found |= self.add_coverage_hist()                                                                                        │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/coverage_hist.py", line 21, in add_coverage_hist │
│     s_name, data_by_phenotype = parse_fine_hist(f)                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/coverage_hist.py", line 140, in parse_fine_hist  │
│     cum_pct = cum_cnt / total_cnt * 100.0                                                                                            │
│ ZeroDivisionError: division by zero                                                                                                  │

The underlying problem may be that this specific sample z04748 is a complete dropout. That is, the FASTQ files of that sample are 20 byte long, meaning completely empty.
Our FASTQ->BAM pipeline automatically ran through the samplesheet and built data for this sample, where of course a number like total_cnt is 0, triggering the division-by-zero error.

File that triggers the error

z04748_syw1.wgs_fine_hist.csv

MultiQC Error log

/data/NGS/NovaSeq_analysis_data/220729_A00866_0397_BHCMYCDMXY_syw1_bam> multiqc -f . /data/NGS/NovaSeq_analysis_data/220729_A00866_0397_BHCMYCDMXY_syw1_fastq/

  /// MultiQC 🔍 | v1.12

|           multiqc | Search path : /data/netapp/DE_MHG2_FASCL01_fvol3/NGS/NovaSeq_analysis_data/220729_A00866_0397_BHCMYCDMXY_syw1_bam
|           multiqc | Search path : /data/NGS/NovaSeq_analysis_data/220729_A00866_0397_BHCMYCDMXY_syw1_fastq
|         searching | ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 5001/5001  
|          mosdepth | Found 79 reports
╭───────────────────────────────────────────── Oops! The 'dragen' MultiQC module broke... ─────────────────────────────────────────────╮
│ Please copy this log and report it at https://github.com/ewels/MultiQC/issues                                                        │
│ Please attach a file that triggers the error. The last file found was: ./z04748_syw1/z04748_syw1.wgs_fine_hist.csv                   │
│                                                                                                                                      │
│ Traceback (most recent call last):                                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/multiqc.py", line 651, in run                                   │
│     output = mod()                                                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/dragen.py", line 64, in __init__                 │
│     samples_found |= self.add_coverage_hist()                                                                                        │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/coverage_hist.py", line 21, in add_coverage_hist │
│     s_name, data_by_phenotype = parse_fine_hist(f)                                                                                   │
│   File "/home/helfert/miniconda3/lib/python3.8/site-packages/multiqc/modules/dragen/coverage_hist.py", line 140, in parse_fine_hist  │
│     cum_pct = cum_cnt / total_cnt * 100.0                                                                                            │
│ ZeroDivisionError: division by zero                                                                                                  │
│                                                                                                                                      │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
|            fastqc | Found 79 reports
|            fastqc | Sample had zero reads: 'z04748_syw1'
|        bclconvert | 2 lanes and 79 samples found
|           multiqc | Compressing plot data
|           multiqc | Deleting    : multiqc_report.html   (-f was specified)
|           multiqc | Deleting    : multiqc_data   (-f was specified)
|           multiqc | Report      : multiqc_report.html
|           multiqc | Data        : multiqc_data
|           multiqc | MultiQC complete

[cmatKhan]

I just had the same error. This is from a minimal test set which seems to be a bit too minimal.

  ╭───────────────── Oops! The 'rseqc' MultiQC module broke... ──────────────────╮
  │ Please copy this log and report it at                                        │
  │ https://github.com/ewels/MultiQC/issues                                      │
  │ Please attach a file that triggers the error. The last file found was:       │
  │ ./4/hg002_T1_star.bam_stat.txt                                               │
  │                                                                              │
  │ Traceback (most recent call last):                                           │
  │   File "/usr/local/lib/python3.10/site-packages/multiqc/multiqc.py", line 65 │
  │     output = mod()                                                           │
  │   File "/usr/local/lib/python3.10/site-packages/multiqc/modules/rseqc/rseqc. │
  │     n[sm] = getattr(module, "parse_reports")(self)                           │
  │   File "/usr/local/lib/python3.10/site-packages/multiqc/modules/rseqc/bam_st │
  │     d["unique_percent"] = (float(d["mapq_gte_mapq_cut_unique"]) / t) * 100.0 │
  │ ZeroDivisionError: float division by zero                                    │
  │                                                                              │
  ╰──────────────────────────────────────────────────────────────────────────────╯

The bam stat file looks like this:

#==================================================
#All numbers are READ count
#==================================================

Total records:                          0

QC failed:                              0
Optical/PCR duplicate:                  0
Non primary hits                        0
Unmapped reads:                         0
mapq < mapq_cut (non-unique):           0

mapq >= mapq_cut (unique):              0
Read-1:                                 0
Read-2:                                 0
Reads map to '+':                       0
Reads map to '-':                       0
Non-splice reads:                       0
Splice reads:                           0
Reads mapped in proper pairs:           0
Proper-paired reads map to different chrom:0

hg002_T1_star.bam_stat.txt

[ewels]

I just had the same error.

I don't think so, your error is from the RSeQC module, whereas the first is from Dragen. Whilst they both are raising ZeroDivisionErrors the bugs are quite separate. That is a very standard core Python exception type to raise.

Thanks for commenting and providing test data though, I'll check and try to fix both bugs 👍🏻 I missed labelling this issue before so hadn't caught it before release, apologies.

Phil

[ewels]

@stefanhelfert I just tried your example data with the latest release and I don't get the same error any more. So I think that this was fixed as part of one of the several large Dragen PRs that have been merged recently. I correctly get a report with two basically empty plots, showing zero coverage.

@cmatKhan - I found and fixed the RSeQC bug in 8c3fff2. Thanks for reporting! 🎉