Oct 11 - "Function, design and evolution of gene circuitry"

Michael Savageau, Biomedical Engineering, UC Davis

Abstract: The physical basis for complex phenotypes is the context-dependent expression of the organism's genome. The context is provided by the life cycle of the organism; the molecular mechanisms of gene regulation interpret that context and orchestrate appropriate responses. The regulation of many gene systems has been studied in detail, and the results have revealed an enormous diversity of molecular elements and circuits. We are just beginning to understand the functional implications of such variations in design and to grasp the factors that have influenced their evolution. The relationship of these variations in design to the phenotype of the organism is even less clear. A quantitative systems approach is required to elucidate these relationships, for without it our understanding will remain descriptive and lack predictive value. This talk will focus on such a theory that relates molecular mechanisms of gene control to the organism's natural environment. Three aspects will be emphasized: mathematical methods for characterizing and comparing the function of gene circuits, biological design principles that have been revealed by such an approach, and implications for the evolution of a specific, well-studied gene circuit. The results provide surprising predictions concerning the organism's phenotype and habitat. The lessons learned provide a deeper understanding of intact biological systems in their natural environment. They also provide guidance as we attempt to re-engineer gene circuits with the intent of correcting pathologies through rational treatment or of producing useful products through biotechnology.

Oct 18 - "The chaotic escape of ultracold atoms from an optical trap -- A study in fractals"

Kevin Mitchell, Natural Sciences, UC Merced

Abstract: I will discuss recent theoretical predictions on the escape of small packets of ultracold and Bose-condensed atoms from an atomic trap. The underlying trap dynamics is chaotic, which profoundly influences the escape rate of atoms from the trap. In particular, fractal patterns are predicted to be directly observable in the escape time data, with sufficient resolution to accurately resolve several fractal layers. Much of the talk will address the intricate nature of these fractals and how they can be accurately described and classified using new tools in symbolic dynamics. These tools are applicable to a wide range of chaotic escape phenomena.

Oct 25 - "Optimal design of spatial distribution networks"

Michael Gastner, Santa Fe Institute

Abstract: We consider the problem of constructing facilities such as hospitals, airports, or malls, in a country with a non-uniform population density, such that the average distance from a person's home to the nearest facility is minimized. We review some previous approximate treatments of this problem which indicate that the optimal distribution of facilities should have a density that increases with population density, but does so slower than linearly, as the two-thirds power. We confirm this result numerically for the particular case of the United States with recent population data using two independent methods, one a straightforward regression analysis, the other based on density-dependent map projections. We also consider strategies for linking the facilities to form a spatial network, such as a network of flights between airports, so that the combined cost of maintenance of and travel on the network is minimized. We show specific examples of such optimal networks for the case of the United States.

Nov 1 - "Applications and algorithms for dynamic semantic graphs"

Tina Eliassi-Rad, Lawrence Livermore National Laboratory

Abstract: Dynamic semantic graphs are directed graphs that evolve over time. The term "semantic" refers to the fact that such graphs have multi-modal nodes and multi-relational links. Dynamic semantic graphs are mostly used for representing a consistent view of data from disparate time-evolving sources and can grow to very large sizes. These graphs are used in various applications involving pattern matching, pattern discovery, and anomaly detection, just to name a few. In this talk, I will provide examples of real-world dynamic semantic graphs, their properties, applications, and algorithmic challenges arising from their size and embedded semantics.

Nov 8 - "Learning in Neural Networks and the Brain"

Tony Bell, Redwood Center for Theoretical Neuroscience, UC Berkeley

Abstract: not available

Nov 15 - "Software networks and the biology of polymorphic design"

Chris Myers, Cornell Theory Center, Cornell University

Abstract: Within large software systems lurk complex networks of interacting components. One of the principal elements of software design is the decision about how to organize such networks, that is, how to decompose functional units at the appropriate level of granularity, and to structure interactions among those units. Much of the complexity of large software networks arises from the need for systems to be at once both functional and evolvable. These related yet distinct goals seem to conspire to produce characteristic software network topologies. Perhaps not surprisingly, these topologies are reminiscent of those found in the study of other complex, evolving networks arising in biology, sociology, and other fields of technology. The connection to biology is especially intriguing, as biological systems are also faced with the need to be both functional and evolvable (as well as robust to noise and perturbation, which software systems are typically not). In this talk, I will first describe software networks, their topology, and the relation of that topology to software design. Then I will sketch some connections between software design and biological organization, focusing primarily on the role that promiscuity and polymorphism play in negotiating tradeoffs between specificity and control, and in satisfying needs for function, robustness and evolvability.

Nov 29 - "The role of network analysis in elucidating gene function and evolution"

Alpan Raval, Keck Graduate Inst for Applied Biosciences

Abstract: Genes and proteins interact in many ways to form complex networks that are amenable to graph-theoretic analyses. We show that the protein-protein interaction network in yeast, while possessing a power-law degree distribution, actually admits various "scales" that enable one to identify different classes of proteins based on their network properties alone. This classification further results in the identification of functionally homogenous sub-networks and thus provides clues towards predicting the function of unannotated proteins. We also show that protein interaction network properties can be used to predict synthetically lethal pairs of genes in yeast with reasonable accuracy. Finally, we discuss how a combined analysis of seven putative predictors of evolutionary rates of yeast genes, including two network properties, reveals a single dominant predictor for evolutionary rate that is linked to the number of translation events.

Dec 6 - "Measuring and managing distributed networked systems"

Chen-Nee Chuah, Electrical and Computer Engineering, UC Davis

Abstract: As the Internet becomes an essential part of our everyday life, it has grown to a complex distributed networked system that is hard to characterize. This talk highlights the needs to develop the foundations for measuring and validating the system behavior of the Internet. For example, how does one measure, predict, or validate end-to-end reachability or security property between two points in the network?

The first part of the talk focuses on characterizing service availability of IP-networks, an important metric that captures transient routing dynamics and reflects user-perceived network performance. We illustrates how graph-theoretic properties like node degree or network diameter) fails predict service availability, making it imperative to use our new metrics for comparing topologies and network design.

The second part of the talk discusses a methodology for validating end-to-end reachability of packets, which requires the knowledge of not only the physical network topology, but also the software configuration of various components along the way, such as routers, firewalls, and NATs. The concatenation of the configuration rules of routers, firewalls, NATs, can be viewed as a specialized software program. In our initial work, we applied static analysis to examine firewall rules for policy violations and inconsistencies at different levels: intra-firewall, inter-firewall, and cross-path. The results are promising, paving a new direction to extend theories of single-machine computation to predict and/or validate end-to-end behavior of large, complex networks.