<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2743.600" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>
<DIV><FONT face=Arial size=2>Presentación de Tesis de Licenciatura en Ciencias
de la Computación</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=Arial size=2> “Learning
Robust Dynamics in Bacterial
<BR> Regulatory Networks
Using GENIE”</FONT></DIV>
<DIV> </DIV><FONT face=Arial size=2>
<DIV><BR>Alumno: Patricio Traverso<BR>Director:
Igor Zwir</DIV>
<DIV> </DIV>
<DIV>Jurados: Pablo Jacovkis, Alberto Kornblihtt , Irene
Loiseau</DIV>
<DIV> </DIV>
<DIV>Jueves 2 de septiembre, 15:30 hs, aula a confirmar.</DIV>
<DIV> </DIV>
<DIV>Están todos invitados.</DIV>
<DIV> </DIV>
<DIV><BR>ABSTRACT:</DIV>
<DIV> </DIV>
<DIV>One of the big challenges of the post genomic era is determining<BR>when,
where and for how long genes are turned on or off. Gene<BR>expression is
determined by protein-protein interactions among<BR>regulatory proteins and with
RNA polymerase(s), and protein-DNA<BR>interactions of these trans-acting factors
with cis-acting DNA<BR>sequences in the promoters of regulated genes. These
interactions<BR>define complex genetic networks, which designs have
motivated<BR>researchers to draw direct analogies with established
techniques<BR>in electrical engineering. As with the construction of
electrical<BR>circuits, the gene circuit approach uses mathematical
and<BR>computational tools in the construction and posterior analysis of<BR>a
proposed network diagram. So far, the qualitative agreement<BR>between model and
experiment in a series of studies depends both<BR>on the design of the network
topology, which most of the times<BR>includes uncertain connections between
genes, as well as on the<BR>dynamic behavior of the network, which is affected
by the<BR>ambiguity inherent to the biological processes (e.g., monomer
or<BR>dimer binding of promoters, enzymes having kinase and/or<BR>phosphatase
activities, etc.) and the mathematical models used to<BR>represent them (e.g.,
Boolean or continuous models; reverse or<BR>forward algorithms). Moreover,
the number of genes considered in<BR>the networks is usually large compared to
the number of the<BR>available measurements (e.g., time-point expression), thus,
more<BR>than one possible model may be consistent with the subjacent
data.<BR>Finally, the data always contains a substantial amount of
noise,<BR>provided by the systematic variability of the experiments, which<BR>in
addition to previous problems, makes it difficult to deduce
the<BR>implications of the underlying logic of genetic networks
through<BR>experimental techniques alone.</DIV>
<DIV> </DIV>
<DIV>We propose a methodology, termed GENIE for Gene Expression<BR>Networks
Iterative Explorer, that embraces the uncertainty<BR>inherent to the biological
problem and the imprecision of their<BR>underlined mathematical models by using
an iterative approach.<BR>First, GENIE proposes a network topology based on DNA
sequence<BR>analysis of transcription factor interactions, which,
together<BR>with previous knowledge from the literature, constitute the
raw<BR>material for the architecture design. Second, we transform
the<BR>hypothesis provided by the network topology, by means of its<BR>possible
chemical reactions and physical constraints, into a<BR>system of nonlinear
ordinary differential equations, whose<BR>variables are concentrations of
proteins, mRNA, etc. Rather than<BR>advocating a single, definitive model
of the genetic network, we<BR>describe a variety of models that have different
strengths,<BR>weaknesses and domains of applicability. Third, the network
models<BR>are evaluated by testing the ability to reproduce the
behavior<BR>observed in vivo of their subjacent non-linear models,
each<BR>characterizing the time-dependent change in concentration of
the<BR>components, including kinase, phosphatase, and
transcription<BR>activities. Fourth, the successful models are tested
by<BR>considering different emergent systemic properties, such as<BR>flexibility
to reproduce all possible functional patterns, and<BR>robustness to changes in
parameters and initial conditions. Fifth,<BR>we revisit the original topology
and iterate, developing adaptive<BR>models of genetic networks. Finally, a
decision making process<BR>reveals the most realistic models, which are examined
by a<BR>datamining process providing substantial insights from the
modeled<BR>genetic systems.</DIV>
<DIV> </DIV>
<DIV>We applied GENIE to uncover regulatory networks in the enteric<BR>bacteria
Salmonella enterica serovar Typhimurium by focusing on<BR>the PhoP/PhoQ and
PmrA/PmrB two-component systems, which govern<BR>virulence and the adaptation to
low Mg2+ and high Fe3+<BR>environments, respectively. The study of the
PhoP regulon<BR>constitutes a special challenge due to the multiplicity
of<BR>PhoP-controlled targets, and the connectivity of the PhoP/PhoQ<BR>system
with other two-component systems, such as PmrA/PmrB,<BR>transcriptional
regulators, and alternative RNA polymerase sigma<BR>factors. We verified
our predictions by investigating<BR>transcription and PhoP and PmrA protein
binding to the identified<BR>promoters in vivo.</DIV>
<DIV> </DIV>
<DIV><BR></FONT> </DIV>
<DIV><FONT face=Arial size=2>Igor Zwir<BR>Howard Hughes Medical
Institute<BR>Department of Molecular Microbiology, Box 8230<BR>Washington
University School of Medicine<BR>St. Louis, MO 63110-1093 USA<BR>Phone, lab:
(314)-362-3691<BR>FAX: (314)-747-8228<BR>email: <A
href="mailto:zwir@borcim.wustl.edu">zwir@borcim.wustl.edu</A><BR><A
href="http://www.microbiology.wustl.edu/dept/postdoc/zwir">http://www.microbiology.wustl.edu/dept/postdoc/zwir</A></FONT></DIV></BODY></HTML>