GLaD: A Mixed-Membership Model for Heterogenous Tumor Subtype Classification

This page contains supplemental data files and computer code for GLaD. The manuscript and supplementary information can be accessed at the Bioinformatics website here.

Code

We provide a python module containing the core functionality of GLaD. This module depends on the following modules: numpy, scipy, itertools, h5py tempfile, logging and time. To run in multithreaded mode, you also need the multiprocessing module.

Download GLaD

To check the functionality of the module simply call the module from the command line using $ python glad.py

Data

The simulation data used for the experiment in section 4.1 are provided in HDF5 and CSV format.

• Sample Size = 30 [HDF5] [CSV]
• Sample Size = 300 [HDF5] [CSV]
• Sample Size = 3000 [HDF5] [CSV]

The tissue mixture experiments in Section 4.2 are available from GEO as GSE19830. The data used for this experiment was downloaded on 2013-06-07.

The GBM data from the TCGA project used in Section 4.3 is available here. We downloaded and used the unified, scaled, filtered data: unifiedScaledFiltered.txt.