[TYPES/announce] Postdoc or PhD position available

[Cedric Lhoussaine]

A POSTDOC position for 2 years, or a PhD position for 3 years, in
computational biology at the Université de Lille 1, is available
starting in September 2009. This project will held at the LIFL computer
lab (http://www.lifl.fr) and the Interdisciplinary Research Institute
(http://iri.ibl.fr) in Lille and within the bioComputing group
(http://www.lifl.fr/BioComputing).

Candidates should have a background in bioinformatics. Additional
knowledge in computer science and/or biology such as modeling and
simulation, gene regulation, semantics of programming languages
(CAML, ...), compilation, lambda-calculus, pi-calculus, typing, logic,
model checking will be appreciated. 

The proposed research project is summarized below.

-- 
Cedric Lhoussaine
LIFL, UMR 8022 CNRS, USTL Batiment M3,
59655 Villeneuve d'Ascq Cedex FRANCE
Phone (IRI): +33 (0)3 62 53 17 09
Phone (Lifl): +33 (0)3 28 77 85 70
Fax: +33 (0)3 62 53 17 90
Email: Cedric.Lhoussaine at lifl.fr
Web: www.lifl.fr/~lhoussai




TITLE
A Uniform Approach for Stochastic Modeling with Spatial Aspects in
Systems Biology

SUMMARY
Experimental techniques in biological research are increasingly
sophisticated, and allow to accumulate increasingly sharp knowledge.
However, the understanding and the representation of cells and living
matter remain fragmented, and do not report all interactions and
interdependencies between biological mechanisms. Systems Biology tries
to overcome this problem. Its long-term objective is to provide the
theoretical and practical tools for describing, studying and predicting
the behavior of biological systems. To this end, it is necessary to
integrate into formal models the increasing and various experimental
data available until now. This formalization subsequently allows the
simulation and the qualitative analysis of models, and thus the
observation and the in-silico study of the properties of biological
systems. The predicted behaviors may themselves be validated
experimentally, if necessary.

Much effort within Systems Biology concentrates on gene regulatory
networks. Essential genomic functions are strongly influenced by low
number of biological actors, causing stochastic phenomena. The knowledge
in this domain remains widely insufficient despite of novel experimental
techniques, that allow to collect data from single cells. For example,
recent work suggested that the non-uniform spatial distributions of
elements in eukaryotic nucleus plays an important role in gene
regulation. Today, it is impossible to unravel the genomic functions of
space solely by experimental methods. It is thus necessary to develop
modeling tools and simulation to study such systems in silico.

An established computational approach for the modeling of gene networks
consists in considering the cell as an executable entity, that is a
system of concurrent agents with programmable interactions. Thus, a
model is a concurrent program and simulating the model means to execute
the program. This approach was initiated by A. Regev and E. Shapiro
within the π-calculus. It was followed by many others, using process
algebras extended with stochastic control. Such approaches benefit from
solid theoretical foundations and formal, unambiguous, semantics that
permit the study and the proof of properties of systems.

However, the pi-calculus approach remains difficult to use for
biologists who are not familiar with object-centered programming, that
is in which the interaction capabilities are described for every agent
independently of other. That is why we are currently focusing on a more
intuitive alternative which preserves the expressiveness of concurrent
and stochastic control. This alternative, commonly called ruled-base
programming, is based on the rewriting of multi-sets of parametrized
terms. It remains intuitively close to chemical reactions which are
familiar biologists.

In this project, we propose to investigate the design and implementation
a novel stochastic modeling language. Its goal is to allow to address
any type of gene network with a concurrent control, that depends on the
position of biological actors. The way positions are described should
remain independent of any type of spatial data. We hope to end with a
unified framework, accessible to biologists, which extends on existing
rule-based approaches while proposing compartments (with variable
volume). The existing checking and prediction methods will be extended
to this new language. We shall apply our language to specific case
studies of cellular biology with spatial aspects in eukaryotic gene
regulation: position of chromosomes in the nucleus, appearance and
preservation of compartments, cross-talk between chromosomes, etc.