A comparative analysis on computational methods for fitting an ERGM
to biological network data
Sudipta Saha
Munni Begum
Dalla Lana School of Public Health, University of Toronto, Toronto, ONM5S 2J7, Canada
Department of Mathematical Sciences, Ball State University, Muncie, IN47306, USA
Network Biology
ISSN 2220-8879
http://www.iaees.org/publications/journals/nb/online-version.asp
2015
5
1
1
12
International Academy of Ecology and Environmental Sciences
Hong Kong
16 October 2014
25 November 2014
1 March 2015
biological networks
regulatory networks
exponential random graph models
Monte Carlo
maximum likelihood estimation
maximum pseudo likelihood estimation
E. coli.
Exponential random graph models (ERGM) based on graph theory are useful in studying global biological network structure using its local properties. However, computational methods for fitting such models are sensitive to the type, structure and the number of the local features of a network under study. In this paper, we compared computational methods for fitting an ERGM with local features of different types and structures. Two commonly used methods, such as the Markov Chain Monte Carlo Maximum Likelihood Estimation and the Maximum Pseudo Likelihood Estimation are considered for estimating the coefficients of network attributes. We compared the estimates of observed network to our random simulated network using both methods under ERGM. The motivation was to ascertain the extent to which an observed network would deviate from a randomly simulated network if the physical numbers of attributes were approximately same. Cut-off points of some common attributes of interest for different order of nodes were determined through simulations. We implemented our method to a known regulatory network database of Escherichia coli (E. coli).
DOI 10.0000/issn-2220-8879-networkbiology-2015-v5-0001
http://www.iaees.org/publications/journals/nb/articles/2015-5(1)/computational-methods-for-fitting-an-ERGM-to-biological-network-data.pdf