|
|
|
Mardi 21 Septembre - Max ostrowski and Sven Thiele (Universität Potsdam, Germany) |
|
|
Written by Pierre PETERLONGO
|
" Solving the Network Reconstruction Problem" and "Metabolic Network Expansion with Answer Set Programming"
Mardi 21 Septembre 2010 - 10H30 Salle Aurigny
Max ostrowski: Solving the Network Reconstruction Problem
Models of biological systems are of high scientific interest and
practical relevance.
A common approach for creating such systems is to construct
descriptive models from series of experiments.
This manual process usually starts from a model defined using existing
biological knowledge which is then gradually refined by appeal to data
gathered in subsequent experiments.
The resulting models are merely consistent with the gathered
experimental data, and, besides simulation,
no true indication can be given how well the resulting model captures
the
biological system.
In the field of Automatic Network Reconstruction many approaches have
been studied that automate the reconstruction of models from experiment
data.
However, most of these systems are heuristic driven or rely on further
assumptions.
We continue an approach from Marwan et. al. that computes a complete
list of
all minimal models that are in accord with the given experiments.
Given all models, further experiments can be planned to discriminate and
reduce the number of possible models.
We tackle the problem using Answer Set Programming, which combines the
computational power through a highly efficient inference engine with
the
simple modeling language of logic programs to describe the problem.
We therefore use off-the-shelf tools for state-of-the-art Boolean
constraint solving.
Sven Thiele: Metabolic Network Expansion with Answer Set Programming
We propose a qualitative approach to elaborating the biosynthetic capacities
of metabolic networks.
In fact, large-scale metabolic networks as well as measured datasets suffer
from substantial incompleteness.
Moreover, traditional formal approaches to biosynthesis require kinetic
information, which is rarely available.
Our approach builds upon a formal method for analyzing large-scale metabolic
networks.
Mapping its principles into Answer Set Programming (ASP) allows us to address
various biologically relevant problems.
In particular, our approach benefits from the intrinsic incompleteness-tolerating capacities of ASP.
Our approach is endorsed by recent complexity results, showing that the reconstruction of metabolic networks and related problems areNP-hard.
|
|
