Title of internship: Abstract Interpretation of Metabolic Networks for Improving Gene Knockouts Predictions

Keywords: Systems Biology, Synthetic Biology, Metabolic Networks,
Abstract Interpretation, Gene Knockout, Modeling

Research Group: BioComputing, LIFL (CNRS UMR 8022), University of Lille

Supervision: Mathias John, Mirabelle Nebut, Joachim Niehren (HDR).

Place: LIFL, M3 building

Continuation: A successful candidate could continue with a PhD project
in the BioComputing group on this topic.

Stipend: About 420€/month

Topic:

In the area of synthetic biology, a major branch of research is to
genetically modify ordinary bacteria in order to turn them into
efficient factories for valueable ressources, such as fuel or medicals
[4]. Since 2 years, the BioComputing group of the LIFL started a
collaboration with the ProBioGEM group at Polytech’ Lille to develop
gene knockout strategies for the production of lipo-peptides in
Bacillus subtilis based on formal modeling and computational analysis
[2,3]. To this end, the major challenge turned out to be that models
largely lack kinetic information, as they are simply not known for
most of the processes occurring in cells.

Abstract interpretation [1] is an approach for the static analysis of
programs based on the approximation of variable domains. In prior
work [2], the BioComputing group developed an approach to predict gene
knockout strategies based on abstract interpretation. The main idea is
to consider models in terms of sets of chemical reactions and to
approximate domains in order to reason about the effects of reaction
deletions indendent of unavailable kinetic information. Thereby, the
major challenge is to provide approximations that are on one hand
effective, i.e. they integrate as much information as possible in
order to ensure the quality of results, and on the other hand
feasible, i.e. they do not require information that is
unavaible. Based on the developed approach first gene knockout
strategies have been predicted that were proven to be effective in
wet-lab experimentation.

Currently, the bioComputing group looks to extend this work in order
to refine the developed abstractions so that they are capable of
integrating novel quantitative information. Preliminary results on a
first more fine-grained abstract interpretation have already been
developed.

Objective: Continue the preliminary work about more fined-grained
abstract interpretations.

Expectations: Decent knowledge on fundamental and practical aspects of
computer science research would be helpful, in combination with
interdisciplinary interest in biology and chemistry (no biological
knowledge necessary).

References: (see also http://www.lifl.fr/BioComputing/#Publications)

[1] Patrick Cousot and Radhia Cousot. Systematic design of program
analysis frameworks. In POPL, pages 269–282, 1979.

[2] Mathias John, Mirabelle Nebut, Joachim Niehren. Knockout
Prediction for Reaction Networks with Partial Kinetic Information. To
appear in the VMCAI proceedings, 13th International Conference on
Verification, Model Checking, and Abstract Interpretation.

[3] François Coutte, Mathias John, Max Béchet, Mirabelle Nebut,
    Joachim Niehren, Valérie Leclère, Philippe Jacques.  Synthetic
    Engineering of Bacillus subtilis to Overproduce Lipopeptide
    Biosurfactants.  9th European Symposium on Biochemical Engineering
    Science, Sep 2012, Istanbul, Turkey.

[4] Jay D. Keasling. Manufacturing Molecules Through Metabolic
Engineering. Science, 330(6009):1355– 1358, December 2010.