Efficiently run Augustus in multi-core environments

parallel_augustus cuts the input genome in chunks to feed to Augustus. The number of chunks and Augustus processes launched in parallel are configurable.

Installation

pip install --user parallel_augustus

Requirements

• Python >= 3.8
• Augustus

Execution

Before launching parallel_augustus, please make sure that Augustus is available in your path. If you run augustus -h and do not encounter an error, you are good to go.

parallel_augustus first divides the genome (-g argument) in the desired number of chunks (-c argument, it can be any number > 1). After that it launches -p processes of Augustus in parallel until there is no chunks left. You can pass parameters to Augustus via the --extra flag. At least a --species=thing is required by Augustus.

IMPORTANT: the --extra argument should be the last one in the command line due to shortcomings in the argparse python module.

With all that in mind, a typical parallel_augustus command line will look like this:

parallel_augustus -g genome.fasta \
    -o augustus_results \
    -c 500 \
    -p 20 \
    --extra '--species=human --gff3=on'

This command creates the output directory augustus_results, divides the genome into 500 chunks and launches 20 processes of Augustus in parallel. At the end, results are concatenated into augustus_results/augustus.gff.