Warnings

/!\: Act with care; this workflow uses significant memory if you increase the values in .masterconfig. We recommend keeping the default settings and running a test first.

/!\: For now workflow only tested with SPN generation

/!\: For now dont run multiple split at once

How to Use

A. Running on the CBIB

1. Set up

Clone the Git repository and switch to my branch:

git clone https://forgemia.inra.fr/pangepop/MSpangepop.git
cd MSpangepop
git checkout dev_lpiat

2. Add your files

• Add a .fasta.gz file; examples can be found in the repository

3. Configure the pipeline

• Edit the .masterconfig file and the visor_sv_type.yaml in the .config/ directory to suit your needs.
• Edit the job.sh and the ./config/snakemake_profile/clusterconfig.yaml with your email.

4. Run the WF

The workflow has two parts: split and simulate. Always run the split first and once its done (realy quick) run the simulate. To do so run the following commands:

sbatch job.sh [split, simulate] [dry, dag]

If no warnings are displayed, run:

sbatch job.sh [split, simulate] 

Nb 1: to create a visual representation of the workflow, use [dag]. Open the generated .dot file with a viewer that supports the format.

Nb 2: Frist execution of the workflow will be slow since images need to be pulled.

B. Run localy

• Ensure snakemake and singularity are installed on your machine.
• Modify the .masterconfig file and visor_sv_type.yaml in the .config/ directory as needed.

./local_run [split, simulate] [dry, dag]

If the workflow cannot download images from the container registry, install Docker, log in with your credentials, and rerun the workflow:

docker login -u "<your_username>" -p "<your_token>" "registry.forgemia.inra.fr" 

Workflow

More informations

The variants generation is inspired by VISOR.

Each variant type has a size distribution file (bins = 100 bp) in folder sv_distributions. The data was extracted from An integrated map of structural variation in 2,504 human genomes (Sudmant, et al. 2015). The distributions are used to randomly sample each structural variant size.

You can extract a VCF from the graph using the vg deconstruct command. It is not implemented in the pipeline.

Dependencies

TODO pandas, msprime, argprase, os, multiprocessing, yaml, Bio.Seq singularity, snakemake vg:1.60.0, bcftools:1.12, bgzip:latest, tabix:1.7.