↑ Return to Vision

Biology

In biology, we advanced a completely new hypothesis stating that biochemical networks are probabilistic computers capable of performing Bayesian inferences.

This contrasts with traditional descriptive models that assume that variability and multiplicity in cell signaling pathways are either not significant, or simply spurious result from specific biological constraints. In our view, these are key features of functional values, which should be analyzed and reinterpreted in term of probabilistic computation.

In the past decades, numerous works have shown that complex behaviors in humans and other primates can be better understood in the framework of Bayesian modeling. However the abilities to perceive objects, to communicate or to make rational decisions in spite of incomplete and uncertain knowledge are thought to be reserved to highly developed brains. We will challenge this assumption, and we will attempt to show that even unicellular organisms could perform some elementary probabilistic reasoning. A partner of the consortium (ULG) is working on the unicellular algae Chlamydomonas.

A picture of our experimental model, Chlamydomonas reinhardtii. It is a well-studied mobile unicellular microalgae with a fully decrypted genome (Merchant et al, 2007), and a complex metabolism. A better understanding of the short-term and long-term adaptation of this kind of organisms to variations of energy input might have a considerable impact in biology, and beyond, for use them as bio fuel.

A picture of our experimental model, Chlamydomonas reinhardtii. It is a well-studied mobile unicellular microalgae with a fully decrypted genome (Merchant et al, 2007), and a complex metabolism. A better understanding of the short-term and long-term adaptation of this kind of organisms to variations of energy input might have a considerable impact in biology, and beyond, for use them as bio fuel.

We proposed some interpretation relating some general principles governing biological signal processing at various scales (macromolecule, cell, organism) to  the basic principles of probabilistic computation.

For this purpose we focusssed on the molecular and intracellular scales, in contrast with current approaches considering that information processing occurs at neuronal or cortical scales.

We described some  elementary biochemical processes, e.g. messenger diffusion, configuration changes in allosteric macromolecules, cascades of reactions, etc. and their equivalents in terms of elementary probabilistic operators. We built the equivalent probabilistic model from the knowledge of real cell signaling networks.

We studied  the response of perturbed biological systems at different time scales. This analysis was based on new genomic and metabolic data processing of Chlamydomonas mutants’ response to variations of energy inputs.

From this analysis, we have developped  important theoretical models of biological systems conceived as huge finite state machines, constrained by thermodynamic laws, but able to adapt to their changing environments.