A workflow (job) engine/pipeline for bioinformatics and scientific computing.
Project description
pypipegraph
Build status: | |
---|---|
Documentation | https://pypipegraph.readthedocs.io/en/latest/ |
Code style |
Introduction
pypipegraph: is an MIT-licensed library for constructing a workflow piece by piece and executing just the parts of it that need to be (re-)done. It supports using multiple cores (SMP) and (eventually, alpha code right now) machines (cluster) and is a hybrid between a dependency tracker (think 'make') and a cluster engine.
More specifically, you construct Jobs, which encapsulate output (i.e. stuff that needs to be done), invariants (which force re-evaluation of output jobs if they change), and stuff inbetween (e.g. load data from disk).
From your point of view, you create a pypipegraph, you create jobs, chain them together, then ask the pypipegraph to run. It examines all jobs for their need to run (either because the have not been finished, or because they have been invalidated), distributes them across multiple python instances, and get's them executed in a sensible order.
It is robust against jobs dying for whatever reason (only the failed job and everything 'downstream' will be affected, independend jobs will continue running), allows you to resume at any point 'in between' jobs, and isolates jobs against each other.
pypipegraph supports Python 3 only.
30 second summary
pypipegraph.new_pipeline()
output_filenameA = 'sampleA.txt'
def do_the_work():
op = open(output_filename, 'wb').write("hello world")
jobA = pypipegraph.FileGeneratingJob(output_filenameA, do_the_work)
output_filenameB = 'sampleB.txt'
def do_the_work():
op = open(output_filenameB, 'wb').write(open(output_filenameA, 'rb').read() + ", once again")
jobB = pypipegraph.FileGeneratingJob(output_filenameB, do_the_work)
jobB.depends_on(jobA)
pypipegraph.run()
print('the pipegraph is done and has returned control to you.')
print('sampleA.txt contains "hello world"')
print('sampleB.txt contains "hello world, once again")
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for pypipegraph-0.197-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 289b257b7afb83c5da38b6467403a76139696d4f5cd03a23b1b50d091e4b47a9 |
|
MD5 | 49341e1afe1b8bcfb71650c0f63b624c |
|
BLAKE2b-256 | 25811475601d682a572df173a89529255910c545c79ff54030c722e9ed2966ce |