[Pl-seminar] Aviv Regev/CGR Special Seminar: Thursday 11 April at 10:00 am

[Mitchell Wand]

A little far from our usual stuff, but I saw the words "pi calculus"
in the abstract...  --Mitch 

------- start of forwarded message (RFC 934 encapsulation) -------
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Return-Path: <owner-colloquium@deas.harvard.edu>
Delivered-To: wand@ccs.neu.edu
X-Authentication-Warning: minuet.das.harvard.edu: majordom set sender to owner-colloquium@deas.harvard.edu using -f
X-Mailer: Mozilla 4.7 [en] (WinNT; U)
X-Accept-Language: en
Precedence: bulk
Reply-To: Peter Arvidson <peter@deas.harvard.edu>
From: Peter Arvidson <peter@deas.harvard.edu>
Sender: owner-colloquium@deas.harvard.edu
To: colloquium@deas.harvard.edu
Subject: Aviv Regev/CGR Special Seminar: Thursday 11 April at 10:00 am
Date: Mon, 08 Apr 2002 09:49:54 -0400

Presenter: Aviv Regev 
[Dept. Cell Research and Immunology, Faculty of Life Sciences, Tel Aviv
University, Tel Aviv, and Department of Computer Science and Applied
Mathematics, Weizmann Institute of Science, Rehovot, Israel]
Date: Thursday, Apr.11
Time: 10:00 
Place: Fairchild 177

Presentation Title: Biological Networks 2.0: Modeling, Homology, and
Reconstruction

Abstract:

Biochemical processes, carried out by networks of proteins, mediate the
interaction of cells with their environment and are responsible for most
of the information processing inside cells. Recently, much interest has
been focused on system level studies of such networks. Here, we present
two complementary computational approaches: building modular predictive
models of known pathways and reconstructing novel ones from
high-throughput gene expression data. 

We developed modular predictive models for biochemical networks using
the pi-calculus process algebra, which was originally developed for
describing concurrent computational processes. We extended the original
language to account for key biological aspects - stochasticity,
dynamics, modularity and compartmentalization - and developed the BioSPI
system for the representation and stochastic simulation of biochemical
pathways. Our approach has two major benefits. The modular nature of the
calculus allows incremental modeling of complex networks and alternation
between different levels of complexity. For example, we investigated the
modular organization of the circadian machinery and its cross-regulation
of the cell cycle, and proposed a new model for the gating effect
exerted by the circadian clock on cell division in rapidly dividing
cells. A second exciting direction from our work is a study of the
evolution of molecular processes based on a homology of processes. This
unique measure of homology is derived from a general pi-calculus
framework for comparing the behavior of computational processes 

To discover novel pathways, we took a complementary approach, attempting
to reconstruct pathways from experimental data. We analyzed gene
expression profiles with the probabilistic graphical framework of
Bayesian networks, and learned a finer structure of interactions between
genes, including causality, mediation, activation and inhibition
relations. We devised methods to reconstruct significant sub-networks of
interacting genes from these relations. We applied this approach to
expression data from S. cerevisiae mutants, and uncovered a plethora of
binary and tertiary relations.  These represent both known and novel
biochemical, regulatory and functional links, many of which are
context-specific and could not be identified by other methods. Based on
these features, we automatically reconstructed several sub-networks
integrating metabolic, signaling and regulatory relations into
biologically-coherent structures. 


1 Joint work with Ehud Shapiro, Eva Jablonka, William Silverman, Naama
Barkai (modeling), Dana Pe'er, Gal Elidan and Nir Friedman
(reconstruction).
------- end -------