Skip to content
Deep hierarchical programs combined with Markov random fields.
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
_fig
examples
shm
.flake8
.gitignore
.styles.yapf
LICENSE
MANIFEST.in
README.md
setup.cfg
setup.py

README.md

Structured hierarchical models

Project Status

Deep hierarchical programs combined with categorical Markov random fields.

About

Structured hierarchical models are a family of generative models for probabilistic inference of causal effects from genetic perturbation screens. SHMs utilize classical hierarchical models to represent heterogeneous data and combine them with categorical Markov random fields to encode biological prior information over functionally related genes. The random field induces a clustering of the genes which helps to inform inference of parameters in the hierarchical model. SHMs are designed for extremely noisy, experimental data sets for which the true data generating process is difficult to model.

An SHM has the following basic structure.

The top-level categorical variables $z_g$ are latent cluster assignments for $g$ genes with effect sizes $\gamma_g$. The effect sizes are used to parameterize a deep hierarchical model, for instance to represent a heterogeneous data set $y_{gci}$ of multiple conditions $c$ and interventions $i$.

Installation

You can install the package using:

pip install https://github.com/cbg-ethz/shm/archive/<version>.tar.gz

where <version> is the most recent release on GitHub. Please don't install from the master branch directly.

Citation

If you like this package, please consider citing it:

Dirmeier, S. and Beerenwinkel, N. (2019). Structured hierarchical models for probabilistic inference from perturbation screening data. bioRxiv. https://doi.org/10.1101/848234

Author

Simon Dirmeier simon.dirmeier @ web.de

You can’t perform that action at this time.