Biological inferences

image2_1_retouche
Figure 1. Maps of forces in membrane microdomains, as obtained by our novel inference method [3].

Internship/Phd proposals

Potential and Diffusion Mapping in Lipid Rafts
     Inference methods are a subject of major interest and we apply them to a number of biological problems, in addition to the one detailed for chemotaxis.
In [1], we demonstrate the efficiency of Belief Propagation (BP) methods for the interactions among protein domains from protein-protein interaction data. Constraints resulting from the latter are effectively dealt with by local messages exchanged in the BP method.
     In [2] we address the problem of tracking particles (or any other moving object) in dense conditions, i.e. when ambiguities are present as to which particle matches with whom in successive time snapshots. The problem arises in an extremely wide range of problems, from fluid dynamics to biological systems, e.g. when tracking flocks or dense bacterial colonies. The work in [2] presents preliminary results on methods that we are developing and seem particularly promising to meet the challenge.
     Finally, in [3] we recently developed inference methods to analyze the motion of proteins and lipids in cell membranes. A major physical motivation to the interest in this issue stems from the actively debated origin of membrane compartmentation. Most commonly, motion is followed by using single-molecule (or single-particle) tracking based on labeling the biomolecule of interest with an organic fluorophore or an inorganic nanoparticle and detecting the position of the biomolecule via a signal related to its label (fluorescence, light scattering, etc.). The biomolecule trajectories thus obtained are usually analyzed by plotting the mean-square displacement (MSD) of the molecule as a function of time. By comparing and fitting the MSD curves with analytical behaviors expected for different modes of motion, e.g. free Brownian diffusion, directed, confined or anomalous motion, parameters like diffusion coefficients and, in the case of confined motion, domain sizes are extracted. The usage of inference methods, exploiting all the information hidden in the trajectories and not just the second-order moment, seem particularly fruitful and to allow for detailed and specific maps of the forces acting within membrane microdomains (see Fig. 1), as well as general tools for determining diffusivity in confined regions. We are currently pursuing the application of the novel method on data obtained for the dynamics in neuronal cells by the groups of M. Dahan and A. Triller (ENS Paris).



[1]. Message-passing algorithms for the prediction of protein domain interactions from protein-protein interaction data. M. Iqbal, A.A. Freitas, C.G. Johnson, M. Vergassola, Bioinformatics, 24 2064-70, 2008.

[2] Belief Propagation and Beyond for Particle Tracking, M. Chertkov, L. Kroc, M. Vergassola arXiv:0806.1199v1 2008.

[3]. Inferring maps of forces inside cell membrane microdomains. Masson JB, Casanova D, Türkcan S, Voisinne G, Popoff MR, Vergassola M, Alexandrou A. Phys Rev Lett. 102 048103. 2009.
Featured in Virtual Journal of Biol. Phys. Research – Feb. 1, 2009, Volume 17, Issue 3;
Virtual Journal of Nanoscale Science & Technology – Feb. 9, 2009, Volume 19, issue 6