University of California, Irvine

 

Scientific Inference Systems Laboratory

 

The purpose of the Scientific Inference Systems Laboratory (SISL) is to develop and apply computing technologies to assist scientists in all phases of the scientific process [1] in biology, planetary science and other sciences.  Some relevant technologies include biological modeling and knowledge representation software, biological and geological modeling frameworks, and mathematical methods for probabilistic modeling, nonlinear optimization, and network inference.  SISL is an element of the Institute for Genomics and Bioinformatics and of the Department of Information and Computer Science at the University of California Irvine.

 

Reference

[1] “Machine Learning for Science: State of the Art and Future Prospects”, Eric Mjolsness and Dennis DeCoste, Science 293, 2051-2055, September 14 2001.

 

Personnel

 

Eric Mjolsness, Associate Professor of Information and Computer Science

414 B Computer Science Building, University of California, Irvine 92697-3425, USA

email: emj@uci.edu.  phone: 949 824 3533; fax: 949 824 4056.

Henrik Jonsson, Postdoctoral Scholar, Caltech

 

Classes

 

ICS 280, Seminar in Computational Systems Biology, Winter 2003.

Readings and project in the computational study of transcriptional regulatory networks and signal transduction pathways. (Forerunner to proposed ICS 277C for 2004). 

 

Research Areas

 

Systems Biology Research Area

 

The expression “systems biology” is often used to denote attempts to build a computable, predictive scientific understanding of living systems from an approximate understanding of the behavior of their molecular components such as single genes and proteins.  Often the focus is on systems at the level of pathways and regulatory networks, but analysis at this level has consequences for the understanding of disease, multicellular development, and evolution.  Continuing improvements in laboratory instrumentation and data sources (including genome-scale sequence data, expression data, and many types of imagery) have made it possible and often essential to understand such biological systems in silico – meaning in computer simulation.

 

Specific research directions in systems biology at SISL build on early work in gene regulation network modeling [1] applied to Drosophila development [2].  More recent developmental modeling has added two-way interactions between material properties of tissues and genetic/signaling regulatory networks [3] (figures 1, 2 and 3).  Signal transduction is receiving increasingly detailed modeling treatment [4] to include important phenomena such as scaffold proteins and membrane localization in signal transduction complexes.  Eucaryotic gene regulation is open to much more detailed modeling [5] than in our previous gene regulation network approaches.  These efforts are substantially aided by mathematical model generation software such as the Cellerator reactions-to-model translation system  [4,6] (see figure 4) and, in the future, the open-source SIGMOID pathway modeling database project.

 

 

 

 

 

 

 

 

 

 


Figure 1.  Shoot apical meristem of Arabidopsis thaliana with gene expression patterns for (a,b) CLAVATA3, CLAVATA1 Fletcher et al., Science 283, 1911-1914, (1999), and (c)WUSCHEL.  WUSCHEL expression domain which may be an “organizing center” Brand et al Science 289 (617-619) 2000.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 2. Simulated expression domains using Genetic/Signaling Regulatory Network (GSRN) model.  (a)  Sharply L1-specific factor analogous to  ATML1 in red, WUSCHEL (initial condition) in blue. (b) Diffusely L1-peaked factor analogous to ACR4. (c) CLAVATA1 expression.  Simulation due to Henrik Jonsson.

 

 

 

 

 

 

 

 

 

 

 


Figure 3. Illustration of GSRN model.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 4. Model generation with Cellerator. (a) data flow, (b) ring oscillator example.

 

 

Systems Biology References

 

[1] “A Connectionist Model of Development'', Eric Mjolsness, David H. Sharp, and John Reinitz, Journal of Theoretical Biology, vol 152 no 4, pp 429-454 , 1991.

 

[2] “Model  for Cooperative Control of Positional Information in Drosophila by bcd and Maternal hb”, John Reinitz, Eric Mjolsness and David H. Sharp, Journal of Experimental Zoology 271:47-56,  1995.

 

[3] “Modeling plant development with gene regulation networks including signaling and cell division”, E. D. Mjolsness, H. J¨onsson, and B. E. Shapiro. In Proceedings of the Third International Conference on Bioinformatics of Genome Regulation and Structure (BGRS’2002), 2002.

 

[4] “Automatic model generation for signal transduction with applications to MAP-kinase pathways”, B. E. Shapiro, A. Levchenko, E. Mjolsness. In Foundations of Systems Biology, ed. H. Kitano, MIT Press 2001.

 

[5] “Gene Regulation Networks for Modeling Drosophila Development'”, Eric Mjolsness, in Computational Methods in Molecular Biology, eds.  J. M. Bower and H. Bolouri, MIT Press 2001.

 

[6]  “Developmental simulation with Cellerator”, B. E. Shapiro and E. D. Mjolsness. In Proceedings of the Second Inernational Conference on Systems Biology (ICSB), pp 342–351, 2001.

 

 

Solar System Exploration Research Area

 

Planetary surfaces are rich in complex geological processes and in the resources required to sustain life as we know it.  They provide challenging domains for scientific exploration by orbiting satellites, robots, and human explorers equipped with automated assistants.  Future intelligent space systems will benefit from a built-in understanding of geological and other planetary processes, and will be capable of scientific inference about such processes.  Unlike current spacecraft and missions, future space exploration systems will be deeply affected by biotechnology in their design, fabrication, and support for human explorers.  Space medicine problems such as microgravity-induced muscle atrophy and the health risks of space radiation, currently major obstacles for human missions to Mars, will be understood and mitigated.

 

Specific research areas currently include statistical modeling for geological process inference, causes of microgravity-induced muscle atrophy, and transgenic plant development models for space agriculture.

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 5.  Illustration of Multirover Integrated Science Understanding System simulation [3].

 

References for Solar System Exploration Research Area

 

[1] “Strategies for autonomous rovers and Mars”, Martha S. Gilmore, Rebecca Castano, Tobias Mann, Robert C. Anderson, Eric D. Mjolsness, Roberto Manduchi, and R. Stephen Saunders, Journal of Geophysical Research - Planets, December 25 2000.

 

[2] “The Synergy of Biology, Intelligent Systems, and Space Exploration”, E. Mjolsness and A. Tavormina, IEEE Expert Systems April-May 2000.

 

[3] “An Integrated System for Multi-Rover Scientific Exploration”, Tara Estlin, Tobias Mann, Alexander Gray, Gregg Rabideau, Rebecca Castano, Steve Chien, and Eric Mjolsness, Proceedings of the American Association for Artificial Intelligence conference, July 1999.

 

 

Multiscale Mathematical Methods Research Area

 

Improved mathematical methods are essential to creating new capabilities in scientific inference systems.  Relevant proglem areas include nonlinear optimization [1,2,3], complex statistical models [4], and inverse problems such as circuit or network inference.

 

References for Multiscale Mathematical Methods

 

[1]  “Convergence Properties of the Softassign Quadratic Assignment Algorithm”, Anand Rangarajan, Alan Yuille, and Eric Mjolsness, Neural Computation 11(6), 1455-1474 1999.

 

[2] “Multiscale Optimization in Neural Networks”, Eric Mjolsness, Charles Garrett, and Willard Miranker, IEEE Transactions on Neural Networks, vol 2 no 2 1991.

 

[3] “Algebraic Transformations of Objective Functions”, Eric Mjolsness and Charles Garrett, Neural Networks, vol 3, no 6, pp 651-669, 1990.

 

[4] “Stochastic Parameterized Grammars for Bayesian Model Composition”, E. Mjolsness, M. Turmon, W. Fink, NIPS workshop on Software Support for Bayesian Analysis Systems. Organizers: W. Buntine,  B. Fischer, J. Schumann, December 2000.

 

Opportunities at SISL

 

There may be educational and research opportunities for graduate students, postdoctoral scholars, and undergraduate students at SISL.  If interested, contact Professor Mjolsness by email.

 

 

Internet Resources Relevant to SISL

 

Software

Cellerator biological model generation system

SIGMOID pathway modeling database project

 

Research Groups and Institutions

Machine Learning Systems Group, Jet Propulsion Laboratory, California Institute of Technology

Wold Laboratory, California Institute of Technology

Meyerowitz Laboratory, California Institute of Technology

Signal Transduction and Cell-Cell Communication Laboratory, Johns Hopkins University

Center for Cell Mimetic Space Exploration, University of California, Los Angeles

Systems Biology Software at Keck Graduate Institute

 

Collaborators

Rebecca Castano, Jet Propulsion Laboratory

Dennis Decoste, Jet Propulsion Laboratory

Andre Levchenko, Johns Hopkins University

Elliot Meyerowitz, California Institute of Technology

John Reinitz, SUNY Stony Brook

Bruce Shapiro, Jet Propulsion Laboratory

Barbara Wold, California Institute of Technology

… and others

 

Data

Regulatory interactions

Yeast ChIP-chip gene regulatory network data (October 2002).

General Repository for Interaction Datasets

Biomolecular Interaction Network Database BIND

Transfac

Gene Ontology

Pathway databases:

KEGG pathway database

Enzyme reaction database Brenda

Enzyme Metabolic Pathways EMP

What Is There WIT

Species-specific genomes:

            Yeast: Saccharomyces Genome Database

            Fly: FlyBase

            Worm: WormBase

            Human/Mouse comparison for human chromosome 19