Skip to content

Latest commit

 

History

History
56 lines (44 loc) · 2.7 KB

README.md

File metadata and controls

56 lines (44 loc) · 2.7 KB
Raw

NOTICE: These pipelines scripts have been deprecated. Use "graphtyper genotype" or "graphtyper genotype_sv" instead!

Graphtyper pipelines

This repository has pipeline scripts for older versions of Graphtyper (pre v2.1). They are only here for reproducability of older genotyping runs and since a few publications have references to them. The scripts depends on the following tools:

  • Graphtyper
  • GNU parallel
  • samtools
  • tabix

The pipeline scripts will automatically search for these tools in your PATH directories, but you can also specify their paths by changing the config.sh file or create your own my_config.sh file. You can also set various variables/parameters in the configuration files (which are documented there). The only required variable is the GENOME variable, which you must set to your reference genome FASTA file.

Usage

Short version

Small variants genotyping

bash make_graphtyper_pipeline.sh <BAM> [CONFIG] | bash

where BAM can either be a single BAM file or a file with a list of BAM files. CONFIG is the configuration file to use (default: ./config.sh). The SV genotyping is run similarly:

bash make_graphtyper_sv_pipeline.sh <BAM> [CONFIG]

Long version

Running on a computer cluster

The command in the "short version" is a typical use case for someone who wants to run the Graphtyper pipelines on a personal computer. If you have a computer cluster, you are likely to be using some workload manager. Integrating the pipeline script with your workload manager should be easy, as each line can be run independently and thus can be run in parallel. For example, if your workload manager of choice is Slurm you could use:

bash make_graphtyper_[sv_]pipeline.sh <BAM> [CONFIG] | \
  while IFS='' read -r line
  do
    srun -c 4 $line &
  done

wait

This will run all regions as separate jobs on your cluster using Slurm with four threads allocated.

Failed runs

The pipeline automatically detects which VCF files are missing from the results directory. So if you had some failed runs you can again run:

bash make_graphtyper_[sv_]pipeline.sh <BAM>

to get the commands to run only failed regions. If no commands are outputted, it means all runs have completed. If your runs fail becaue of memory, we recommend decreasing the NUM_SLICES_RUNNING variable to reduce the memory requirement without having any effect on quality.

Example

A test example config can be found in test_config.sh. You can run the test with:

make -f pipelines.make test

If everything is configured properly the test results should appear in test/results/20/000000500-000009499.vcf.gz. That output should be the same as in test/expected.vcf.gz.

License

GNU GPLv3