MANILA, OCTOBER 22, 2007 (STAR) STAR SCIENCE By Eduardo R. Mendoza, Ph.D.  (Second of two parts)

Modeling NetwOrks in the Life and Environmental Sciences (MOLES)

For modeling biochemical networks, MOLES research focuses on methods which preserve the biological semantics and hence allow continued active project participation of the bioscientists. Petri Nets, for example, are very useful in formalizing the qualitative aspects of networks, even when kinetic data for various components are unavailable. BST (Biochemical Systems Theory) or power law formalism has proven effective for constructing models concerned with average changes of system components over time, usually with coupled ordinary differential equations. These “canonical models” have only two kinds of kinetic parameters, which are easily interpreted biologically. The harder task of estimating such parameters from experimental data is rightfully assigned to the modeling partner, but the computational results are easily validated biologically by the experimenter.

Parameter estimation for BST models — in the form of S-Systems or Generalized Mass Action (GMA) models — is a very active field of research, with over 25 original papers published in the last four years on the subject. The paper of Gonzalez et al (4) in “Bioinformatics” (the journal with the highest impact factor in the field) documents the successful SMILES collaboration of computer scientists at UPD and microbiologists at LMU Munich. The paper’s major contribution — the effective application of simulated annealing to parameter estimation — was verified using experimental data of the microbiologists on E.coli’s cadBA system.

A more recent result of Filipino research (5) in this area is part of a growing collaboration with Prof. Eberhard Voit, a leading BST expert from Georgia Tech. The major part of this collaboration is the MAD (Munich-Atlanta-Diliman) initiative, which aims to build (in engineering terminology) a “problem solving environment” for Canonical Modeling by Bioscientists (CaMBio). An important activity toward the “problem solving environment” is a systematic benchmarking of available parameter estimation methods. Currently four UPD researchers and five graduate students are involved in this MAD Benchmarking project within SMILES.

Education with SMILES

From the very beginning, research and education for Computational Systems Biology have gone hand in hand in the SMILES initiative. These efforts over the last four and a half years have led to date not only to five ISI publications (four published, one accepted) and four to five in preparation for submission in the next six months, but also to the establishment of specialization “tracks” within existing MS (Math, Computer Science) and PhD (Math) programs. Five MS students (three Math, two Computer Science) have finished and the first two PhDs are expected to graduate within the next six months. The annual “Biological Structures” course initiated at the Marine Science Institute by Giselle Concepcion and Ed Padlan has also contributed greatly to the students’ success, as it enabled non-biologists to get an overview of important concepts in modern biological research. A “Computational Systems Biology Group,” led by the labs of Henry Adorna and Prospero Naval, was established in March 2007 to further advance research and graduate education at the Computer Science Department.

Lessons from SMILES

The first four and a half years of SMILES have established a Filipino foothold in Computational Systems Biology. The insistence on tight integration of experimental and modeling work was the most important success factor. In this period, due to current lack of appropriate experimental resources in the country, most of the collaboration was done with groups abroad, particularly in Europe. The collaboration was nevertheless effective through the use of novel Internet-based resources (Wikis, Skype videoconferences for project meetings) and organizational discipline (e.g. use of professional software development methodologies, goal-focused meeting structures and follow-ups). In the future, I expect more local collaborations as appropriate infrastructure is built up, e.g. in the National Science Complex. Further development of chip-oriented technology will dramatically drive costs of currently expensive equipment down, so that more quantitative experimental work can be done locally. The presence of knowledgeable computational partners will help in progressing serious systems biology research in the country.

SMILES has also shown the feasibility of interdisciplinary research based on mutually perceived value. To date, the initiative has not had to ask for any funding from Philippine sources. Many of the experimental groups abroad support the computational work through a Research Assistant Program (RAP) because they were impressed by the initial results and the talent of the Filipino students involved. SMILES projects have led to three Max Planck Postdoctoral Fellowships and four PhD scholarships from various sources.

SMILES is a “proof of concept”: similar initiatives will work for other institutions interested in computational life sciences. MBaRC (Manila Bay Research Corridor, pronounced “embark”), to be launched with DLSU Manila, UP Manila and Mapua in February 2008, will build on the lessons from SMILES. In the same month, SMILES celebrates its fifth anniversary by co-hosting the 10th International Conference on Molecular Systems Biology (

* * *

(4) Orland R. Gonzalez, Christoph Küper, Kirsten Jung, Prospero C. Naval Jr., Eduardo R. Mendoza, Parameter Estimation Using Simulated Annealing for S-System Models of Biochemical Networks, Bioinformatics, Vol. 23, No. 4, February 2007

(5) Ricardo C.H. del Rosario, Eduardo R. Mendoza, Eberhard O. Voit, Challenges in Lin-log modeling of Glycolysis in Lactococcus lactis (accepted, under revision for IET Systems Biology)

Chief News Editor: Sol Jose Vanzi

All rights reserved