Science Talk

Stephane Redon · Guest

PhD position: Adaptive molecular dynamics (INRIA - FRANCE)
----------------------------------------------------------

IMPORTANT: APPLICATION DEADLINE MAY 15th, 2008!

An average human cell contains several billion molecules.
Understanding the complexity underlying the mechanisms and
interactions of these molecules would have wide-ranging applications.
For example, gaining a precise understanding of how proteins fold and
interact would have a tremendous impact on drug design.

One way to understand molecular mechanisms is through modeling and
simulation. Unfortunately, the cost of molecular simulations increases
with the size of the simulated systems. This makes it extremely
difficult to study large systems (e.g. viruses) over long time
periods. To address this problem, researchers typically have to
increase the processing power (which might be expensive), or make
arbitrary simplifications to the system (which might bias the study).

We have recently developed an adaptive simulation method [1,2,4,6],
which rigorously and automatically predicts the most mobile groups in
a simulated molecular system, and simulates these groups only. This
method allows the user to adapt the simulation to available resources,
while guaranteeing that a good approximation is found under the
imposed constraints [3,4]. We now want to generalize the adaptive
simulation theory.

In the current approach, the simulated degrees of freedom are the
torsion angles of a molecule, which describe rotations around covalent
bonds [4]. The first generalization sought is to handle other types of
degrees of freedom (e.g., bond lengths and angles). Especially, we
want to generalize the adaptive simulation theory to rigid body
joints, in order to adaptively refine simulations of groups of
molecules.

A second generalization is to handle kinematic cycles in molecules. In
the current theory, the kinematic graph of a molecule has to be
acyclic. This allows us to simulate proteins and the flexibility of
the amino-acids side chains, but may be insufficient to model other
types of biological molecules (for examples, sugars, DNA, etc.). To
generalize the adaptive theory to cycles, new algorithms have to be
designed.

A third generalization is to adaptively handle restraints in molecular
systems (e.g. inter-atomic distances restraints, symmetry restraints,
etc.). These restraints are necessary to help biologists design and
manipulate models of complex, macro-molecular systems.

If time allows, other extensions will be explored. For example, a
challenging task in macro-molecular simulation is the efficient
sampling of the conformational space available to the system. Other
possible generalizations may thus include (adaptive) multiple time
step methods, hybrid Monte-Carlo simulations, etc.

The PhD candidate will develop the mathematics of the generalized
adaptive simulation theory, will design and implement the
corresponding algorithms, and will validate the work on various
biological applications, in collaboration with bio-physicists and
biologists at CEA [5].

Skills and profile
------------------
A strong background in computer science (data structures and
algorithms, programming), mathematics (multivariate analysis, linear
algebra, etc.) and/or physics (mechanics, electrostatics, etc.) is
required. Basic knowledge and taste for biology and nanoscience is a
plus.

Environment
-----------
This position is offered at the =93Grenoble Rhone-Alpes=94 Research Centre
of INRIA, located near Grenoble and Lyon. The Unit counts more than
500 people, within 26 research teams and 10 support services.
The starting date will be between september 2008 and december 15,
2008.
Duration: 3-years position.
Salary and benefits : 1537 =80 NET / month + full health insurance and
social benefits included - salary will be upgraded to 1620 =80 NET /
month the 3rd year.

- INRIA Team: Nano-D (http://nano-d.inrialpes.fr/)
- Research theme: Computational methods for modeling and simulation of
nano-objects
- Leader: Stephane Redon (stephane.redon@inria.fr)
- PhD advisors:
- Stephane Redon - INRIA, Nano-D team - http://nano-d.inrialpes.fr/~redon
- Serge Crouzy (bio-physicist, HDR) - CEA, LBMC
- Location: INRIA Grenoble - Rh=F4ne-Alpes Research Center - France

To apply
--------

Apply online at:
http://www.inria.fr/travailler/mrted/en/doc/details.html?LG=3DEN&nPostingTar=
getID=3D5630

IMPORTANT: APPLICATION DEADLINE MAY 15th, 2008!

Bibliography
------------

(see also http://nano-d.inrialpes.fr/~redon)

[1] S. Redon and M. C. Lin. "An Efficient, Error-Bounded Approximation
Algorithm for Simulating Quasi-Statics of Complex Linkages".
Proceedings of ACM Symposium on Solid and Physical Modeling, 2005.

[2] S. Redon, N. Galoppo and M. C. Lin. "Adaptive Dynamics of
Articulated Bodies". ACM Transactions on Graphics, 25(3), 2005.

[3] S. Morin and S. Redon. "A Force-Feedback Algorithm for Adaptive
Articulated-Body Dynamics Simulation". Proceedings of IEEE
International Conference on Robotics and Automation, 2007.

[4] http://nano-d.inrialpes.fr/~redon/AMQ

[5] CEA is a French government-funded technological research
organisation (http://www.cea.fr/english_portal).

[6] R. Rossi, M. Isorce, S. Morin, J. Flocard, K. Arumugam, S. Crouzy,
M. Vivaudou, and S. Redon. "Adaptive torsion-angle quasi-statics: a
general simulation method with applications to protein structure
analysis and design". Bioinformatics 2007 23(13):i408-i417 (ISMB/ECCB
2007).