PACTLS: A New Algorithm for Inferring Gene Regulatory Networks from Noisy Data

PACTLS is an iterative algorithm for reverse engineering biological interaction networks from both steady-state and time-course experimental data, based on dynamical systems. It was developed by Dr. Francesco Montefusco at the College of Engineering, Mathematics and Physical Sciences of the University of Exeter, U.K., under the supervision of  by Prof. Declan G. Bates, and by Dr. Carlo Cosentino at the Biomechatronics Lab, directed by Prof. Francesco Amato, of the Magna Græcia University of Catanzaro, Italy.

The algorithm explicitly takes into account the effects of measurement noise by employing the Constrained Total Least Squares (CTLS) technique, an extension of the widely used least squares approach, which optimally deals with the presence of correlated noise in the measurements, method previously developed by Dr. Jongrae Kim, Department of Aerospace Engineering at the University of Glasgow, U.K., and used to identify the parameters of differential equation-based systems biology models from noisy data.

PACTLS can also exploit prior knowledge of specific functional associations in the network. Moreover, it uses an edge selection heuristic based on mechanisms underpinning scale--free networks generation, i.e. network growth and preferential attachment (PA), within the reconstruction process, mimicking the evolution of biological networks.

All the software provided on this page is free for non commercial use and the source code can be downloaded here.

Documentation

The software documentation is included in the source code packages available in the download section.

Relevant Publications

The PACTLS algorithm is described in the following companion paper

[1] F. Montefusco, C. Cosentino, J. Kim, F. Amato, D.G. Bates, Reconstruction of Partially-Known Biomolecular Interaction Networks from Noisy Data, submitted to International Journal of Robust and Nonlinear Control, Special Issue on System Identification for Biological Systems, October 2010  (An early version of this paper was submitted to an invited session on ``Modeling and Identification in Systems Biology: Advances and Challenges'' to be held at the IFAC World Congress on Automatic Control, Milano, Italy, 2011).

Other publications by our groups regarding convex optimization and regression methods to infer biological networks

• F. Montefusco, C. Cosentino, F. Amato, CORE–Net: Exploiting Prior Knowledge and Preferential Attachment to Infer Biological Interaction Networks, IET Syst. Biol. 4(5) 2010, 296-310.

• F. Amato, C. Cosentino, F. Montefusco, Inferring Gene Regulatory Networks with a Partially Known Scale-Free Topology, European Control Conference 2009 (ECC’09), Budapest, Hungary, August 23-26, 2009.

• J. Kim, D. G. Bates, I. Postlethwaite, P. Heslop-Harrison, K. -H. Cho, Linear time-varying models can reveal non-linear interactions of biomolecular regulatory networks using multiple time-series data, Bioinformatics 2008, 24:1286-1292.

• F. Amato, C. Cosentino, F. Montefusco, Inferring Scale-Free Networks via Multiple Linear Regression and Preferential Attachment, Proc. of the 16th Mediterranean Conference on Control and Automation 2008 (MED’08), Ajaccio, Corsica, June 25-27, 2008.

• J. Kim, D.G. Bates, I. Postlethwaite, P. Heslop-Harrison, K.H. Cho, Least-squares methods for identifying biochemical regulatory networks from noisy measurements, BMC Bioinformatics 2007, 8(8).

• C. Cosentino, W. Curatola, F. Montefusco, M. Bansal, D. di Bernardo, F. Amato, Linear Matrix Inequalities Approach to Reconstruction of Biological Networks, IET Syst. Biol., Vol. 1, no. 3, pp. 164–173, May 2007.

• C. Cosentino, W. Curatola, M. Bansal, D. di Bernardo, and F. Amato, Piecewise Affine Approach to Inferring Cell Cycle Regulatory Network in Fission Yeast, Biomedical Signal Processing and Control Vol. 2, pp. 208–216, 2007.

Application Examples

PACTLS has been statistically validated by testing it over a set of in silico networks. The numerical dataset and the code for generating and simulating the in silico networks are available in the dowload section.

A biological case-study is presented in [1], dealing with the reconstruction of a cell cycle gene regulatory subnetwork in S. cerevisiae. The dataset and the gold-standard network to be inferred are available in the download section.

The performance of the PACTLS algorithm was compared with a reverse-engineering method based on Bayesian networks (BN), by using the software BANJO

More information about these tests are reported in [1]
 

Requirements

PACTLS has been developed in Matlab® R2007b and needs some routines included in the Optimization and Control System Toolboxes.

Download

Important: This software is distributed only for non-commercial purposes and only for academic use - Commercial users please contact us.

• Supplementary material for [1]:

  • Source code to perform the reverse engineering tests reported in [1], with PACTLS and BANJO on both the in silico and the in vitro data sets. In order to run the tests with BANJO, it is required to download and install the software from the  authors' web site.

Single packages download

• PACTLS ver. 1.0 package including Matlab® source code;

• Source code to generate in-silico networks;

• The in silico data set used in [1] can be downloaded in  two formats:

  • Matlab^® files (for use with PACTLS);

  • Plain text files (for use with BANJO).

• Data set for the biological case study tackled in [1]:

  • Matlab^® files (for use with PACTLS);

  • Plain text files (for use with BANJO).

Contacts

If you have questions/comments/suggestions, please do not hesitate to contact us.

Dr. Carlo Cosentino

Assistant Professor of Systems and Control Theory

Department of Experimental and Clinical Medicine

Magna Graecia University of Catanzaro

v.le Europa, Campus Salvatore Venuta

88100 Catanzaro

Italy

Tel: +39-0961-369-4051

Fax: +39-0961-369-4090

e-mail: carlo.cosentino at unicz.it