/scalable_multiGGM

Associated code for Shaddox E, Stingo F, Peterson CB, Jacobson S, Cruickshank-Quinn C, Kechris K, Bowler R and Vannucci M. (2018). A Bayesian approach for learning gene networks underlying disease severity in COPD. Statistics in Biosciences. 10(1): 59-85.

Primary LanguageMATLAB

scalable_multiGGM

Author: Elin Shaddox

The Matlab files provided for Bayesian inference of multiple graphical models are associated with the following publication:

  • Shaddox E, Stingo F, Peterson CB, Jacobson S, Cruickshank-Quinn C, Kechris K, Bowler R, Vannucci M. (2018). A Bayesian approach for learning gene networks underlying disease severity in COPD. Statistics in Biosciences. 10(1): 59-85. [pdf]

This work improves the computational scalability of our previous work on Bayesian inference of multiple Gaussian graphical models by utilizing a continuous shrinkage prior. This enables the current method to scale to >100 nodes.

These scripts rely on code provided with the following prior publications:

  • Peterson CB, Stingo FC, Vannucci M. (2015) Bayesian inference of multiple Gaussian graphical models. Journal of the American Statistical Association. 110(509): 159—174.
  • Wang H. Scaling it up: stochastic search structure learning in graphical models. Bayesian Analysis. 10 (2015): 351-377

Please cite these publications if you use this code. Thanks!

OVERVIEW OF FILES

Example_multiple_graphs_SSVS.m

Basic example of running the MCMC sampler and generating results summaries on a simple setting with 3 groups with identical dependence structure

MCMC_multiple_graphs_SSVS_final.m

Code for running the MCMC sampler

calc_mrf_C.m

Helper function for calculating the normalizing constant for the MRF prior

generate_sim1_input.m

Script to generate matrices similar to those used as input to the first simulation