vortex_light

Installing locally and running from local installation

1. Clone the repo from GitHub.

git clone https://github.com/zellerlab/vortex_light.git

2. Create a conda environment with NextFlow, e.g. by using the provided environment.yml.

cd vortex_light
conda env create -f environment.yml
conda activate vortex_light

3. Make a copy of the config/run.config file and adjust it to your environment.

4. Run the pipeline

nextflow run /path/to/vortex_light/main.nf --input_dir /path/to/input_files --output_dir /path/to/output_dir -c /path/to/run.config

Note: Nextflow itself requires at least 5GB of memory.

Running from GitHub

This requires a local nextflow installation. If you don't have one, see Steps 1/2 above.

1. Make a local copy of the run configuration file and adjust to your environment.

2. Run the pipeline

nextflow run zellerlab/vortex_light --input_dir /path/to/input_files --output_dir /path/to/output_dir -c /path/to/run.config

Note: Nextflow itself requires at least 5GB of memory.

Input parameters

• --input_dir should be a folder with bam files or with gzipped fastq files. For fastq files, individual samples should be separated into individual folders.
• --output_dir is vlight_out in the local directory by default.
• --skip_<analysis>, --run_<analysis> skips, resp. explicitly requires execution of the specified analysis (pathseq, base_counts (read counts post pre-processing), kraken2)
• --publishMode allows to switch between various modes of how results files are placed in the output_dir (cf. NextFlow documentation)

Notes

• kraken2 can only run when the parameter kraken_database is set.
• pathseq can only run when the parameter pathseq_database is set.

5. Outputs

The output folder contains:

• a subdirectory per sample (named <sample>) with
  • the kraken2 report <sample>.kraken2_report.txt
  • the library size <sample>.libsize.txt
  • pathseq output
    • <sample>.pathseq.bam
    • <sample>.pathseq.bam.sgi
    • <sample>.pathseq.score_metrics
    • <sample>.pathseq.scores

Note that by default, all files in the output folder are symlinks into the work dir! Before you delete the work dir, ensure you have dereferenced copies. Alternatively, change the --publishMode parameter to copy or link (if the target file system supports hard links).