The First Annual RECOMB Satellite Workshop on Systems Biology
The Second Annual RECOMB Satellite Workshop on Regulatory Genomics

December 2nd-4th, 2005.  University of California, San Diego

RECOMB SATELLITE CONFERENCE ON SYSTEMS BIOLOGY

Day 1: Friday DEC 2, 2005

Time	Presentation title	Speaker
7.45 - 8.45	Registration/ Continental Breakfast
8.45 – 9.00	Introductory remarks	Trey Ideker, UCSD
KEYNOTE INVITED TALKS:
9.00 – 9.45	Multivariate Cue/Signal/Response Analysis of Cell Decision Processes	Douglas Lauffenburger, MIT
9.45 – 10.30	An Integrated Approach to the Reconstruction of Molecular Networks	Richard Karp, UC Berkeley
10.30 – 11.00	Break/ Networking opportunities
11.00 – 11.45	Gene Networks for Drug Target Gene Discovery	Satoru Miyano, University of Toyko
11.45 – 12.30	Dynamic modeling of yeast cell stress response	Edda Klipp, Max Planck Institute for Molecular Genetics
12.30 – 2.00	Lunch/ Networking opportunities
2.00 – 2.45	Trees and Forests: How do molecular networks accommodate change?	Aviv Regev, Harvard University
ABSTRACT PRESENTATIONS:
2.45 - 3.45	An Interactive Map of Regulatory Networks of Pseudomonas aeruginosa Genome The Pathalyzer: a tool for visualization and analysis of signal transduction pathways Decomposition of Overlapping Protein Complexes: A Graph Theoretical Method for Analyzing Static and Dynamic Protein Associations	Weihui Wu, Yongling Song, Shouguang Jin, and Su-Shing Chen. David L. Dill, Merrill A. Knapp, Pamela Gage, Carolyn Talcott, Patrick Lincoln, and Keith Laderoute Elena Zotenko, Katia S. Guimarães, Raja Jothi, and Teresa M. Przytycka.
3.45 – 4.00	Break/ Networking opportunities
4.00 – 4.45	Genome-wide discovery of modulators of transcriptional interactions in B lymphocytes Comparison of protein-protein interaction confidence assignment schemes	Kai Wang, Ilya Nemenman, Nilanjana Banerjee, and Andrea Califano Silpa Suthram, Tomer Shlomi, Eytan Ruppin, Roded Sharan, and Trey Ideker
KEYNOTE INVITED TALK:
4.45 – 5.30	An Analysis of Gene Regulatory and Protein Networks in Halobacteria	Leroy Hood, Institute for Systems Biology

RECOMB SATELLITE SESSION ON DEVELOPMENTAL BIOLOGY

Day 2: Saturday DEC 3, 2005

Time	Presentation title	Speaker
8.45 – 9.00	Introductory remarks	David Gifford, MIT
KEYNOTE INVITED TALKS:
9.00 – 9.45	Embryonic Stem Cell Regulatory Networks	David Gifford, MIT
9.45 – 10.30	Functional properties of the gene regulatory network for early sea urchin development	Eric Davidson, California Institute of Technology (CalTech)
10.30 – 11.00	Break/ Networking opportunities
11.00 – 11.45	Systems Biology of Aging	Stuart Kim, Stanford University Medical Center
11.45 – 12.30	Automated Gene Expression Profiling in C. elegans with Continuous Single Cell Resolution	Robert Waterston, University of Washington
12.30 – 2.00	Lunch/ Networking opportunities
ABSTRACT PRESENTATIONS:
2.00 – 4.00	Characterization of the effects of TF binding site variations on gene expression. Towards predicting the functional outcomes of regulatory SNPs Improvement of Computing Times in Boolean Networks Using Chi-square Tests Genome-wide Identification of Regulatory DNA Sequence Elements: A Steganalysis-based Approach	Michal Lapidot and Yitzhak Pilpel Haseong Kim,, Jae K. Lee and Taesung Park Guandong Wang and Weixiong Zhang
4.00 – 4.30	Break/ Networking opportunities
KEYNOTE INVITED TALKS:
4.30 – 5.15	Genome regulatory network controlling gastrulation of the Drosophila embryo	Mike Levine, UC Berkeley
5.15 – 6.00	Industry Round Table: What is the significance of Systems Biology and Regulatory Genomics to industry?	Mel Kronick and Annette Adler, Agilent; Patrick Warren and Janette Jones, Unilever Keith Elliston, Genstruct
6.00	Poster Session (Networking event)

RECOMB SATELLITE CONFERENCE ON REGULATORY GENOMICS

Day 3: Sunday DEC 4, 2005

Time	Presentation title	Speaker
8.45 – 9.00	Introductory remarks	Eleazar Eskin, UCSD
KEYNOTE INVITED TALKS:
9.00 – 9.45	Coactivator and Corepressor Networks Integrated Transcriptional Response Programs	Michael G. Rosenfeld, UCSD
9.45 – 10.30	Regulatory network discovery and evolution	Manolis Kellis, MIT
10.30 – 11.00	Break/ Networking opportunities
11.00 – 11.45	Flexible Arrays - Opportunities and Pitfalls	Mel Kronick, Agilent
ABSTRACT PRESENTATIONS:
11.45 – 12.45	Backup by paralogs decouples genes lethality from centrality: evidence for preferential backup of hubs Causal inference of regulator-target pairs by gene mapping of expression phenotypes Phenotypic divergence correlates with translational control signals in protein coding sequences in yeast species	Ran Kafri and Yitzhak Pilpel David Kulp and Manjunatha Jagalur Orna Man and Yitzhak Pilpel
12.45 –	Closing Comments and Business Meeting Lunch (for follow-up events)

ABSTRACTS: (In order of appearance)

Douglas Lauffenburger, Massachusetts Institute of Technology (MIT):

• Cell behavioral functions are governed by biomolecular networks which translate stimulatory cues (e.g., ligand/receptor binding interactions, mechanical stresses, other physical or chemical insults) into intracellular signals which then influence transcriptional and post-transcriptional, metabolic, and cytoskeletal processes that integrate to effect proximal and ultimate cell responses.  While there is an increasingly effective body of work enhancing our understanding of how intracellular signals are generated by stimulatory cues, an exceptionally difficult challenge is to understand how these signals operate to influence cell behavioral responses.  We are undertaking an effort address this question by means of a combination of quantitative, dynamic protein-centric experimental manipulations and measurement with a spectrum of computational mining and modeling approaches.  Particular application problems include cell migration, differentiation, and death.  This talk will present some overview and a specific example vignette outlining our efforts.

Richard Karp, UC, Berkeley

• Our approach combines broad-scale network analysis with fine-scale pathway reconstruction, supported by large-scale data acquisition and focused experiments. At the network analysis level we combine data about protein-protein interactions, protein-DNA interactions, genetic interactions and gene expression to identify coherently interacting sets of proteins, explore the architecture of protein networks and compare the structure of protein networks of different organisms. At the pathway reconstruction level we plan to employ single-cell flow cytometry measurements of protein activity, as well as ChIP-chip measurements and gene expression data, to reveal the causal relationships and detailed logic of regulatory pathways. We shall discuss algorithmic techniques, required by this approach, for counting different types of subgraphs in a protein interaction network, finding coherent sets of proteins conserved in two or more species, and constructing optimal Bayes nets according to a likelihood criterion. Joint work with Roded Sharan, Ron Shamir, Trey Ideker, Eran Halperin, Mani Narayanan and Joseph Dale.

Satoru Miyano, Human Genome Center, Institute of Medical Science, University of Tokyo

• We developed a series of computational methods based on Bayesian networks for mining gene networks from microarray gene expression data. We combined the Bayesian network approach with nonparametric regression, where genes are regarded as random variables and the nonparametric regression enables us to capture from linear to nonlinear structures between genes. In order to improve the biological accuracy of estimated gene networks, we made a general framework by extending this method so that it can employ genome-wide other biological information such as sequence information on promoter regions, protein-protein interactions, protein-DNA interactions, and subcelluar localization information. By definition, Bayesian network assumes a directed acyclic graph as its underling structure. However, gene regulatory networks may involve feedback loops which violate the acyclicity condition of Bayesian network. In order to resolve the acyclicity restriction of Bayesian network model, we also developed the dynamic Bayesian network with nonparametric regression for time-course gene expression data. Though the problem of finding an optimal Bayesian network is known computationally intractable, we developed an algorithm for searching optimal and suboptimal Bayesian networks in feasible time for small networks. Computational experiments with this search algorithm have provided evidences of the biological rationality of our computational strategy.

These computational methods for estimating gene networks were applied for searching drug target genes. For a given drug, our strategy assumes two kinds of microarray gene expression data: One is a short time-course gene expression data for the drug response. The other is a set of gene expression data obtained by knock-downs of several hundreds of carefully selected genes (one knock-down for each microarray measurement). With these gene expression data, our computational method produces a gene network expressed as a Bayesian network that most strongly relates to the mode-of-action of the drug in cells.

We prepared 270 novel gene knock-downs for HUVEC and the fenofibrate was used as the drug for investigation. Microarray measurements were conducted for these 270 gene knock-downs and the drug responses in time-course. From these data, we generated gene networks of around 1000 genes by using the supercomputer system at Human Genome Center of University of Tokyo. We report an analysis of the computationally estimated gene networks and discuss how we can explore the networks for searching drug target genes, by focusing on the genes around PPAR-alpha, which is known as the agonist of fenofibrate. Along with this talk, we will also mention the computational capabilities and tools that are required for the current and future research.

Edda Klipp, Max Planck Institute for Molecular Genetics

• Investigation of living beings, especially cellular systems, is more and more supported by computational methods like bioinformatics and mathematical modeling, which is an important aspect of Systems Biology (1,2). A frequent method is the description of reaction systems by sets of ordinary differential equations. The structure of the equations is established based on the knowledge about the network structure, i.e. about the relevant pathways and protein-protein interactions, while the parameters are determined from experimental observations, preferentially time course measurements.

Using the power of such models, we investigate stress response processes in a model organism, the yeast Saccharomyces cerevisiae. The adaptation of the cells to environmental changes like nutrient supply, pheromone stimulation (3) or osmotic stress (4) is mediated by signaling pathways that eventually regulate the expression of many genes. The products of such genes, in turn, regulate the metabolism or the cell cycle progression in order to compensate for or adapt to the external stimuli.

As an example, the adaptation of yeast cells to osmotic stress shall be discussed. Cellular osmoregulation covers active processes to monitor and adjust osmotic pressure and to control cell shape, turgor and water content. We combined the experimental investigation of the cellular response to hyperosmotic shock with dynamic mathematical modeling. The model comprises the stimulation of membrane receptors, the subsequent signaling pathway, the activation of gene expression and the adaptation of cellular metabolism to accumulate glycerol, combined with a thermodynamic description of the regulation of volume and osmotic pressure. Model predictions agree well with experimental results obtained under different stress conditions or using certain mutants. Simulations reveal properties of the signaling process and enlighten the roles of different components in the adaptational process. The impact of the activation of the HOG pathway on the progression of cell cycle is also discussed.

The presented examples show that mathematical models are helpful to formulate experimental knowledge in a testable form, to explain hitherto unsolved phenomena and to even predict the outcome of new experiments.

Aviv Regev, Harvard University

Joint work with Amos Tanay and Ron Shamir

• Molecular networks are the information processing devices of cells and organisms, transforming extra- and intra-cellular signals into coherent cellular responses. Networks are also remarkably flexible and can re-configure in an adaptive response to perturbation. Here, we study the evolvability of the yeast transcriptional network, focusing on modules of co-regulated genes. Comparative studies show that regulatory modules are conserved across taxa, but little is known about the mechanisms underlying module evolution. We explored the evolution of cis-regulatory programs associated with conserved modules by integrating expression profiles for two yeast species with sequence data of 15 other fungal genomes. We show that while the regulatory mechanisms accompanying certain conserved modules are strictly conserved, those of other conserved modules are remarkably diverged. In particular, we infer the evolutionary history of the regulatory program governing the ribosomal modules, showing how the infiltration of new cis-elements, the establishment of redundancy in the module's regulatory mechanism, and the loss of other cis-elements, allow the emergence of different regulatory mechanisms that perform similar functional roles. Our results illuminate the dynamics of promoter evolution and may help in understanding the evolvability and increased redundancy of transcriptional regulation in higher organisms. 

Leroy Hood, Institute of Systems Biology:

• I will discuss our large-scale analyses of gene regulatory and protein networks in Halobacteria using the global capture and integration of data from DNA array technologies, genome-wide localization techniques, protein interaction techniques and a variety of other data capture approaches. The resulting models can then be used to formulate hypotheses about specific biological systems and test them by additional rounds of genetic and environmental perturbations. Halobacteria has proved to be a very powerful model system for developing new predictive and integration computational tools—some of which will also be discussed, as will their applications to gene regulatory networks in higher organisms.

Eric Davidson, California Institute of Technology (CalTech)

• This presentation will concern the gene regulatory network (GRN) for endomesoderm specification in sea urchin embryos. The network, which now contains ~50 genes, mainly regulatory genes, is a powerful tool for both explanation and prediction, extending from the phenomenology of development to cis-regulatory functions at the nodes of the GRN. The GRN is formulated on the basis of prior knowledge of the developmental process, detailed observations on spatial and temporal patterns of gene expression, and a large-scale perturbation analysis. The GRN is represented in computational models that permit predictions of cis-regulatory inputs at the nodes of the GRN, and that display the gene regulatory transactions that are active or inactive in given spatial domains of the embryo at given times. Among the computational methodologies developed to support the GRN analysis are a new application of interspecific sequence comparisons which has proven experimentally to be remarkably useful for rapid identification of cis-regulatory elements. Experimental verification of cis-regulatory predictions has been obtained in remarkable detail for many key nodes of the GRN, indicating that it provides a true representation of encoded genomic regulatory logic for early development.

Michael Rosenfeld, University of California, San Diego (UCSD)

Over the past ten years a large network of coactivator and corepressor complexes have been elucidated, and a combinatorial code required in a cell type, developmental, and promoter specific fashion has been described.  The implications of these findings is that there is a temporal order to recruitment of both DNA binding transcription factors and cofactors that is central to the gene activation events in development and homeostasis.  These events permit the cofactor network to serve as sensors for diverse signaling pathways that impact every cell, and permit integrated program of transcriptional response. Combining these observations with contemporary technology of factor location will provide a definition of cohorts of transcription units under similar types of transcriptional control by cis-acting and trans-acting factors.

Manolis Kellis, Massachusetts Institute of Technology (MIT)

Comparative genomic analysis of multiple related species has emerged as a powerful tool for the systematic discovery of biological signals.  In particular, we have developed methods for the de novo discovery of regulatory motifs in complete genomes, by observing their genome-wide conservation across multiple species.  We have applied these methods to multiple complete mammalian and yeast genomes, revealing a global dictionary of regulatory motifs and insights into their function, including transcription initiation, post-transcriptional regulation, and microRNA targeting.  We have also studied the turn-over rate of regulatory motifs across evolutionary time, revealing an underlying birth-death process.  Finally, we study the effects of gene and genome duplication on cellular networks, and the processes governing the emergence of network motifs.

Our results illustrate the power of comparative analyses in the understanding of regulatory networks, their discovery, and their evolutionary dynamics.

Mel Kronick, Agilent

• In the last few years advances in fabrication technology have enabled production of microarrays whose content and format are defined by software without the need for pre-synthesized oligonucleotides or expensive photolithographic masks. This flexibility enables small lot sizes, iterative experimental design and probe optimization, and content and format definition explicitly tailored to the scientific problem of interest. Example applications where these capabilities have already proven valuable include transcript slice variant analysis, comparative genome hybridization, and location analysis (ChIP-on-chip studies). Utilization of this flexibility is not without a cost: unique demands are placed upon array design and data analysis. Technical issues like normalization methods, access to sequence databases, and tiling density requirements get wrapped up in the biological questions. Full utilization of the flexible array will necessitate very close collaboration of molecular biologists with bioinformaticians who can contribute the appropriate computational tools.