Python code used for the Kaggle Connectomics Challenge

Install the dependencies listed in DEPENDENCIES (we recommend using a separate virtualenv)

Then, from the root of the source tree, call

$ python setup.py build_ext --inplace

... and that's it

Our code assumes that the evaluation datasets are stored in PyTables HDF5 files, with each dataset comprising a group containing three arrays named fluorescence, network and network_pos, consisting of the simulated flourescence data, the true set of connections, and the xy positions of the neurons respectively.

The main script used to process an evaluation dataset is basic_workflow.run(). To construct a submission from a pair of validation and test datasets, use basic_workflow.make_submission().

Most of the important core functions are well documented. If you have any other questions, please email alistair.muldal[at]pharm.ox.ac.uk

Copyright 2014 Alistair Muldal