Bayesian Structure Learning in Sparse Gaussian Graphical Models

Abstract

Decoding complex relationships among large numbers of variables with relatively few observations is one of the crucial issues in science. One approach to this problem is Gaussian graphical modeling, which describes conditional independence of variables through the presence or absence of edges in the underly- ing graph. In this paper, we introduce a novel and efficient Bayesian framework for Gaussian graphical model determination which is a trans-dimensional Markov Chain Monte Carlo (MCMC) approach based on a continuous-time birth-death process. We cover the theory and computational details of the method. It is easy to implement and computationally feasible for high-dimensional graphs. We show our method outperforms alternative Bayesian approaches in terms of convergence, mixing in the graph space and computing time. Unlike frequentist approaches, it gives a principled and, in practice, sensible approach for structure learning. We illustrate the efficiency of the method on a broad range of simulated data. We then apply the method on large-scale real applications from human and mammary gland gene expression studies to show its empirical usefulness. In addition, we implemented the method in the R package BDgraph which is freely available at http://CRAN.R-project.org/package=BDgraphComment: Published at http://dx.doi.org/10.1214/14-BA889 in the Bayesian Analysis (http://projecteuclid.org/euclid.ba) by the International Society of Bayesian Analysis (http://bayesian.org/

Structural and parametric uncertainties in full Bayesian and graphical lasso based approaches: Beyond edge weights in psychological networks

“…Bayesian interpretation had already appeared in the original paper introducing lasso [26], for a detailed Bayesian interpretation, see comments from C. Holmes in [27]. Furthermore, full-fledged Bayesian Lasso and Bayesian graphical lasso were also reported in [14], [15], [28]. However, scalable, full Bayesian extension of graphical lasso for discrete data, especially with options for including higherorder interactions is currently not available.…”

Mathematical Modeling of Gene Networks: Review

“…16,17 Parametrik model alternatifleri içinde ise en öne çıkanları Kopula GGM modelidir. 18,19 Bu model, kısaca, GGM ifadesindeki çok değişkenli normal dağılım fonksiyonunun, Gaussian kopula yardı-mıyla yazılmasından türetilmiştir. Diğer yandan verinin özellikle zaman bağlı olması durumunda durum uzay modelleri ve zaman serisi GGM ifadeleri geliştirilmiştir.…”

“…Bahsedilen modellerin tahmininde optimizasyona dayalı yöntemlerin yanı sıra, cezalandırılmış olabilirlik, ters atlamalı markov zinciri Monte Carlo ve doğum-ve-ölüm algoritması gibi parametrik tahmin yöntemleri ve eşik meyil iniş algoritması gibi parametrik olmayan yöntemler sıklıkla kullanılmaktadır. 10,18,19,22 Fakat modelleme ve parametre tahmin yöntemleri halen bu konudaki araştırma başlıklarıdır. Bu nedenle her geçen gün yeni ifadelerin tanımlanması ve uygun algoritmaların geliştiril-mesi konularında, çalışmalar yapılmaya devam edilmektedir.…”

Gibbs Sampling in Inference of Copula Gaussian Graphical Model Adapted to Biological Networks

“…In the study of Dobra and Len [3], this method is used to infer the parameters of the copula Gaussian graphical model (CGGM). In the application of CGGM for the construction of biological networks, it is found that RJMCMC has a long burn-in period [3,4] and additionally, it needs to calculate the Jacobian term for each associated iteration, as in Eq. (2).…”

“…In order to decrease the computational demand and optimize the calculation, Mohammadi and Wit [4] suggest the birth-and-death method. RJMCM with the splitmerge approach have been also intensively studied in the work of Richardson and Green [5].…”