Ph.D.

In the LMD system, Doctorate or Ph.D. is the highest degree awarded by university, and thus by co-option. At least in France, it typically consists in 3 years of research activities, possibly along with teaching ones. It is said that Ph.D. is to learn researching and teaching by the practice. There is no handbook but ideally good mentor to guide you on this journey. Keep in mind, or be prepared if you aim once to start such adventure: the path is not written and if your mentor, as expert of the domain (min 10 years exp.) will suggest you a direction you can rely on, you'll be alone to figure out one way to achieve this goal.

Thanks to a 3 years government grant, I get the opportunity to do a Ph.D. within bio-computing, meaning across 2 domains, biology (mitochondrial physio-pathology) and computer science (multi-agents system simulation) as between 2 universities / laboratories. More horizons, more resources.. simply better ;)

3 years sounds long time, but time flies surprisingly fast:

  • First year, you take your mark, do your publications review to get a more precise picture of the domain, test few potential resolution approach to define the (supposedly) ideal strategy to attack the problem.
  • Second year, you start to accumulate enough results to submit publications, posters or conference speech. Each first time will be a great validation, an exiting challenge as well. Interaction with colleagues is absolutely required: it can bring doubt but most of the time such emulation will boost you to overcome blocking points with unexpected perspectives.
  • Third year, you feel uncertain about the followed path, unsatisfied by the lack of production and already running out of time with thesis writing stage.

These defense. 45 min to summarize 3 years of research. In front of a group of experts who will add 45 min of questions to evaluate the consistency of your results.

What do you get at the end? What are the deliverable items?

Well, out of a title, that you'll may have to explain.. "no no, I"m not a doctor fixing computers..", you published your thesis manuscript (already lazy.. French only), the source code of the simulation (MitoMAS) and eventually other publications.

But by the way, let's dive a bit deeper in the topic, starting with the subject, short but precise enough: "Simulation of mitochondrial metabolism using multi-agents system".

So let's consider your body, composed of cells, each cell composed of different units or organelles with specific functions. One of them are the mitochondria, or in fact the mitochondrial network, kind of "spaghetti plate".

Mitochondria are specific in many ways: having 2 membranes, inner and outer, or even containing their own DNA. Yes, there are 2 kind of DNA: nuclear and mitochondrial. These 2 unique aspects lead to symbiogenesis or endosymbiotic theory as possible origin of mitochondria: external bacteria incorporated during evolution, as a white blood cell (leukocytes) would do but not destroyed this time. Over time, an probably by simplification process, part of mitochondria's DNA migrated / have been replaced by nuclear one, but some proteins are too big to migrate back to internal membrane and typically mitochondrial DNA is still related to these structures.


My thesis focuses on modelling of the inner membrane and mitochondrial enzyme complexes of the respiratory chain embedded in the phospholipid bilayer.

An alternative to the techniques of molecular mechanics representing biological objects at the atomic level models are grains of atoms, “coarse grained” models, allowing to study biological phenomena at time scales of the tenth of a microsecond and space of the order of a tenth of a micron, which corresponds to values of phenomena such as formation of membrane folds.

The proposed model consists in modelling the phospholipids in the form of rigid rectilinear trimers with a implicit solvent. This coarse-grained model gives rise to an agent oriented design implemented as a Multi-Agent System (MAS) called MitoMAS.

Several simulations with MitoMAS hold that the phospholipids hydrophobicity can be modeled with an implicit solvent as evidenced by the appearance of micelles from an initial random mixture. A inter-molecular potential cutoff of 1nm is proving to be a good compromise between realism and effectiveness of simulations. Under certain constraints of lateral pressure, the bi-layer folds are similar to those of the inner membrane. Simulations of heterogeneous systems exhibit the emergence of phospholipid rafts and those of mixed systems phospholipids / proteins confining intra-membrane complexes in small folds.

Prospects are numerous. Besides the exploration "in silico" of new potentials and their parameters, we can consider mixed models such coarse-grained / mesh surfaces to study the interactome of a cell.