This is a re-worked Python implementation of the GeneMANIA algorithm (MANIA is an acronym for M ultiple A ssociation N etwork I ntegration A lgorithm) for predicting gene function using linear regression-based kernel fusion and modified Gaussian field label propagation. The details of the algorithm as well as the design considerations were first published in this 2008 article in Genome Biology.

It operates on affinity networks stored as SciPy sparse matrices. It requires NumPy_ and SciPy_ to function.

This is based on an earlier MATLAB version by myself and Sara Mostafavi, which is available on her website (as well as more up-to-date versions accompanying her recent publications).

This code is released under the 3-clause BSD license.