1: \begin{abstract}

2: Decoding complex relationships among large numbers of variables with relatively few observations is one of the crucial issues in science. 

3: One approach to this problem is Gaussian graphical modeling, which describes conditional independence of variables through the presence 

4: or absence of edges in the underlying graph.

5: In this paper, we introduce a novel and efficient Bayesian framework for Gaussian graphical model determination 

6: which is a trans-dimensional MCMC approach based on a continuous-time birth-death process. 

7: We cover the theory and computational details of the method. 

8: It is easy to implement and computationally feasible for high-dimensional graphs. 

9: We show our method outperforms alternative Bayesian approaches in terms of convergence, mixing in the graph space and computing time. 

10: Unlike frequentist approaches, it gives a principled and, in practice, sensible approach for structure learning. 

11: We illustrate the efficiency of the method on a broad range of simulated data and on large-scale real applications involving gene expressions experiments. 

12: In addition, we implemented the method in the R package {\tt BDgraph} which is freely available at \url{http://CRAN.R-project.org/package=BDgraph}. 

13: 

14: %We then apply the method on large-scale genomic data sets, to show its empirical usefulness.

15: %We then apply the method on large-scale real applications from human and mammany gland gene expressions, to show its empirical usefulness.

16: 

17: %\textbf{Keywords}: Bayesian model selection, Sparse Gaussian graphical models, Non-decomposable graphs, Birth-death process, Markov chain Monte Carlo, G-Wishart.

18: \end{abstract}

19: