"Per Base Sequence Content" in FastQC module can't handle NaN values

[matrulda]

Description of bug:
I discovered that the FastQC plot "Per Base Sequence Content" was empty. All other plots in the FastQC module were generated.

After troubleshooting a bit, I found that the problem was caused by NaN values in fastqc_data.txt. When replacing the NaN values with numbers, the plots were generated as expected.

Would it be possible to add support for NaN values? Or at least print a warning when this happens?

MultiQC Error log:
No error at all.

[2020-07-08 15:40:54,327] multiqc                                            [DEBUG  ]  No MultiQC config found: /lupus/sw/bioinfo/MultiQC/1.9/rackham/lib/python3.7/site-packages/multiqc_config.yaml
[2020-07-08 15:40:54,327] multiqc                                            [DEBUG  ]  No MultiQC config found: /home/matildaa/.multiqc_config.yaml
[2020-07-08 15:40:54,327] multiqc                                            [DEBUG  ]  No MultiQC config found: multiqc_config.yaml
[2020-07-08 15:40:54,327] multiqc                                            [DEBUG  ]  Command used: /sw/bioinfo/MultiQC/1.9/irma/bin/multiqc -v -m fastqc .
[2020-07-08 15:40:54,330] multiqc                                            [DEBUG  ]  Could not connect to multiqc.info for version check: <urlopen error [Errno 113] No route to host>
[2020-07-08 15:40:54,330] multiqc                                            [INFO   ]  This is MultiQC v1.9
[2020-07-08 15:40:54,331] multiqc                                            [DEBUG  ]  Command     : /sw/bioinfo/MultiQC/1.9/irma/bin/multiqc -v -m fastqc .
[2020-07-08 15:40:54,331] multiqc                                            [DEBUG  ]  Working dir : /tmp/troubleshooting_fastqc
[2020-07-08 15:40:54,331] multiqc                                            [INFO   ]  Template    : default
[2020-07-08 15:40:54,331] multiqc                                            [DEBUG  ]  Running Python 3.7.2 (default, Mar 14 2019, 18:52:14)  [GCC 4.8.5 20150623 (Red Hat 4.8.5-36)]
[2020-07-08 15:40:54,331] multiqc                                            [INFO   ]  Searching   : /tmp/troubleshooting_fastqc
[2020-07-08 15:40:54,332] multiqc                                            [INFO   ]  Only using modules fastqc
[2020-07-08 15:40:54,332] multiqc                                            [DEBUG  ]  Analysing modules: fastqc
[2020-07-08 15:40:54,333] multiqc                                            [DEBUG  ]  Using temporary directory for creating report: /scratch/tmpms4c9j77
[2020-07-08 15:40:54,497] multiqc                                            [DEBUG  ]  Ignored 231 search patterns as didn't match running modules.
Searching 2 files..
[2020-07-08 15:40:54,769] multiqc.plots.linegraph                            [DEBUG  ]  Using matplotlib version 3.0.3
[2020-07-08 15:40:54,770] multiqc.plots.bargraph                             [DEBUG  ]  Using matplotlib version 3.0.3
[2020-07-08 15:40:54,876] multiqc.modules.fastqc.fastqc                      [INFO   ]  Found 1 reports
[2020-07-08 15:40:54,936] multiqc                                            [INFO   ]  Compressing plot data
[2020-07-08 15:40:54,970] multiqc                                            [INFO   ]  Report      : multiqc_report.html
[2020-07-08 15:40:54,970] multiqc                                            [INFO   ]  Data        : multiqc_data
[2020-07-08 15:40:54,971] multiqc                                            [DEBUG  ]  Moving data file from '/scratch/tmpms4c9j77/multiqc_data' to '/tmp/troubleshooting_fastqc/multiqc_data'
[2020-07-08 15:40:55,128] multiqc                                            [INFO   ]  MultiQC complete

File that triggers the error:
fastqc_data.txt
Note: I've modified the file a bit: Replacing overrepresented sequences with NNNNNNNNN

MultiQC run details (please complete the following):

• Command used to run MultiQC: multiqc -v -m fastqc .
• MultiQC Version: MultiQC v1.9 (same result when using running in quay.io/biocontainers/multiqc:1.8--py_1)
• Operating System: CentOS Linux release 7.8.2003 (Core) (Uppmax cluster Irma)
• Python Version: 3.7.2
• Method of MultiQC installation: Uppmax module system

Additional context

[ewels]

Thanks for reporting this @matrulda - NaN values are always a pain.

[matrulda]

Yeah, they are. They probably don't occur that often, just for bad runs like this. But even bad runs deserves plots. 🙏

[ewels]

Hi @matrulda,

Finally got to this - should now be fixed in the dev branch on master 👍🏻 Let me know if it works / you run into any problems.