Supported formats

• FASTQ. Only the Sanger variation with a phred offset of 33 and the error rate calculation of 10 ^ (-phred/10) is supported. All sequencers use this format today.

  • Paired end sequencing data is supported.

  • For sequences called by illumina base callers an additional plot with the per tile quality will be provided.

  • For sequences called by guppy additional plots for nanopore specific data will be provided.

• (unaligned) BAM with single reads. Read-pair information is currently ignored.

  • For BAM data as delivered by dorado additional nanopore plots will be provided.

  • For aligned BAM files, secondary and supplementary reads are ignored similar to how samtools fastq handles the data.

Installation

Installation via pip is available with:

pip install sequali

Sequali is also distributed via bioconda. It can be installed with:

conda install -c conda-forge -c bioconda sequali

Quickstart

sequali path/to/my.fastq.gz

This will create a report my.fastq.gz.html and a json my.fastq.gz.json in the current working directory.

To set the directory where the reports are created the --outdir flag can be used. This is useful when using [MultiQC](https://github.com/multiqc/multiqc).

sequali --out-dir /my/dir/all_sequali_reports my.fastq.gz

The html and json filenames can be set separately.

sequali --html before_qc.html --json before_qc.json my.fastq.gz
sequali --html after_qc.html --json after_qc.json my.cutadapt.fastq.gz

Sequali can handle paired-end data.

sequali /sequencing_data/sample100_R1.fastq.gz /sequencing_data/sample100_R2.fastq.gz

Additionally sequali can handle BAM data. Proper pair handling is not yet supported for BAM data, so this is primarily useful for ONT datasets.

sequali /sequencing_data/sample100_dorado_called_hac_v4.30.bam

Sequali by default uses one thread per compressed input file and one thread for the read processing, typically keeping two cores busy. Sequali can also use a single core, which is slower, but typically more efficient for HPC scenarios where multiple files can be run simultaneously. (Below a SLURM example.)

sbatch -c 1 --time 59 --partition short \
--wrap 'sequali --threads 1 /cluster-scratch/myusername/my.fastq.gz'

Using a thread count higher than 2 has no effect. Due to the decompression bottleneck, bringing the full power of multithreading to Sequali has limited utility whilst having a disproportionally high cost in additional code complexity.

For all command line options checkout the usage documentation.

For more extensive information about the module options check the documentation on the module options.

Acknowledgements

• FastQC for its excellent selection of relevant metrics. For this reason these metrics are also gathered by Sequali.

• The matplotlib team for their excellent work on colormaps. Their work was an inspiration for how to present the data and their RdBu colormap is used to represent quality score data. Check their writings on colormaps for a good introduction.

• Wouter de Coster for his excellent post on how to correctly average phred scores as well as the idea for using end-anchored plots from NanoQC.

• Marcel Martin for providing very extensive feedback.

• Agnès Barnabé for creating a Galaxy wrapper.

Citation

If you wish to credit Sequali please cite the Sequali article.

License

This project is licensed under the GNU Affero General Public License v3. Mainly to avoid commercial parties from using it without notifying the users that they can run it themselves. If you want to include code from Sequali in your open source project, but it is not compatible with the AGPL, please contact me and we can discuss a separate license.