genetools: single-cell analysis recipes (work in progress)

Plot gallery

Other features

• Compare clustering results by computing co-clustering percentage.
• Map marker genes against reference lists to find names for your clusters.
• pandas shotrcuts:
  • Split single cell barcodes conveniently.
  • Defensive pandas merging and concatenation methods with strict correctness checks.

Full documentation: https://genetools.maximz.com.

Install

Run pip install --upgrade 'genetools[scanpy]'.

Or if you don't use scanpy: pip install --upgrade genetools.

Usage

To use genetools in a project, add import genetools. Review the documentation and the tests for examples.

Development

Setup:

git clone git://github.com/maximz/genetools
cd genetools
pip install --upgrade pip wheel
pip install -r requirements_dev.txt
pre-commit install

Common commands:

# lint
make lint

# one-time: generate test anndata, and commit so we have reproducible tests in CI
rm -r data
make regen-test-data

# run tests locally
# this is done in a debian-based docker image to ensure image style matches what Github Actions CI will produce
# failing image snapshot tests are recorded in tests/results/
make build-docker-test-image # whenever requirements_dev.txt change
make test

# generate baseline figures (also happens in docker)
make regen-snapshot-figures

# regenerate test data, and baseline figures (also happens in docker)
make regen-test-data

# run tests locally without docker, therefore omitting the snapshot tests
# (the @snapshot_image tests are still executed but the images are not compared. the @pytest.mark.snapshot_custom are skipped altogether.)
make test-without-figures

# docs
make docs

# bump version before submitting a PR against master (all master commits are deployed)
bump2version patch # possible: major / minor / patch

# also ensure CHANGELOG.md updated

CI:

• Main: Github Actions
• Docs: https://app.netlify.com/sites/genetools

Changelog

0.7.6

• Keep core genetools imports working when optional Scanpy helper imports fail.

0.7.0

• Many new stats, plotting, and helper functions. See PRs #24 and #73.

0.6.0

• Two new functions to customize tick labels on any existing plot:
  • wrap_tick_labels: add text wrapping
  • add_sample_size_to_labels: add group sample sizes with a (n=N) suffix
• Make writing PDF figures a deterministic process and make the PDF text editable.
• Scatterplot improvements:
  • Adjust default marker shape and size to work better for most plots.
  • Adjust HueValueStyle so that an explicit marker size is not specified there, only a marker size scaling factor. The scatterplot itself is responsible for defining the base marker size, while the palette of HueValueStyles should be drawable at any marker size.
  • Change legends so that only one marker is drawn to indicate a group's style.

0.5.0 (2022-01-10)

• Improve scatter plots and stacked bar plots.
• Introduce HueValueStyle for granular styling of each hue.

0.4.0 (2020-07-22)

• Centered log ratio (CLR) normalization for Cite-seq protein data.

0.3.0 (2020-06-03)

• Pandas helpers for easier normalization

0.2.0 (2020-06-03)

• Far faster implementation of stats.coclustering
• Introducing helpers.make_slurm_command
• Global submodule import (no longer need to import submodules individually)

0.1.0 (2020-03-06)

• First release on PyPI.