The inference of a disordered free energy potential from the observation of random walks in its landscape.

This work was motivated by the analysis of unzipping data (mechanical opening of the double helics), with the aim to extract information about the sequence of the unzipped molecule. The dynamics of the interface between the open and closed portions of the molecule may be modeled as a random walk in a sequence-dependent landscape. Though single molecule experiments are extremely well controlled, data are noisy due to experimental limitations and to the thermal noise. Sophisticated inference approaches are therefore required to reconstruct the sequence landscape from the observation of one or more realizations of the random walks, and to estimate how many data one should collect for an accurate inference. Our works have so far mainly attracted the attention of the mathematicians who are interested in the inverse problem of random walks in random potentials (works of Adelman and Enriquez, Andreoletti and R. Diel), but we still hope to analyze some new-generation unzipping data (obtained by a double-optical trap).

see the Publication list, Biophysics: Inverse Problems