phlash 1.0.2

phlash is a program for sampling from the posterior distribution of population size history given whole genome sequence data. You can think of it as a Bayesian version of Li & Durbin's popular PSMC program, with a more modern interface and GPU acceleration built in.

Requirements

• Python 3.10 or greater.

Optional but highly recommended:

• An NVIDIA GPU. Any relatively recent model should work. phlash has been tested on:
  • RTX 4090
  • A40
  • A100
  • V100

Installation

phlash can be installed from PyPI using pip:

$ pip install phlash

If your machine contains an Nvidia GPU, you should install the gpu variant in order to enable it. In this case, you should run:

$ pip install \
    -f "https://storage.googleapis.com/jax-releases/jax_cuda_releases.html" \
    --extra-index-url "https://pypi.ngc.nvidia.com" \
    phlash[gpu]

I recommend installing phlash into a separate virtual environment: before running the above, do

$ python3 -mvenv /path/to/phlash  # replace with desired path
$ source /path/to/phlash/bin/activate
$ pip3 install -U pip setuptools  # recent version of pip and setuptools are required

Running the program

See notebooks/example.md for example code and usage instructions.

Troubleshooting / FAQ

I (Jonathan) am happy to assist you with using phlash, as much as my time allows.

• If you encounter a bug (program crash or other unexpected behavior) please file an issue describing the bug.
• If you need help with anything else (installation, running the program, data formatting, interpreting the output, etc.) please open a discussion.

Making Changes & Contributing

Contributions (in the form of Github pull requests) to improve the project are always welcome!

This project uses pre-commit_, please make sure to install it before making any changes::

pip install pre-commit
cd phlash
pre-commit install

It is a good idea to update the hooks to the latest version::

pre-commit autoupdate