Marie Doumic-Jauffret: modelling amyloidosis

Date:
Changed on 25/03/2020
Marie Doumic-Jauffret is one of 4 Inria candidates selected in the young researcher category for the ERC European call for projects 2012. Her project, called Skipper, aims to improve understanding of amyloidosis, a group of diseases that includes prions and Alzheimer’s, in order to help biologists identify therapeutic solutions.

How did you get into research at Inria?

I was not planning to go into research. Some dream of a career in research at a very early age, but that was not my case at all! After studying applied mathematics I did a thesis in partial differential equations applied to lasers. But I wanted a pragmatic job, something to do with what I saw as the "real world". So I decided to look at engineering. That's how I ended up running the engineering department of the VNF (Voies Navigables de France) for 3 years, overseeing the construction of dams and sluice gates.

I learned a lot there. After three years in that position I thought I was ready to do research in the same operational manner. I also realized that my thesis, which seemed rather abstract at the time, had been used by others to conduct laser simulations. In 2007, as I was an engineer at Ponts et Chaussées, I was able to join Inria on secondment. I joined the BANG team, because of its applications in biology which fascinated me.

What is the subject of your ERC Skipper project?

It involves applying a family of equations some members of the BANG team are working on, and that describe the evolution of time among populations, to the modelling of amyloidosis. This category of diseases, such as prion diseases or Alzheimer's, is characterised by deposits of protein aggregates in brain tissue. For a reason we do not fully understand, these proteins change configuration and become capable of polymerisation and of clinging to one other and forming large starch-like aggregates, hence the name amyloid fibrils. This polymerisation process is similar to the mechanisms observed by the team during the division and growth of cells or bacteria, to the extent the polymers grow by adding monomers and divide by splitting. Thanks to technological progress, biologists have accumulated an enormous amount of data that needs to be fully exploited. I hope that my mathematical models can help them by highlighting new data in the scales they already have, that will allow them to take full advantage of these scales and identify those that provide the most information.

This is a multi-disciplinary project. Do you work with biologists?

It is absolutely necessary to ensure the relevance of the model developed. An Inra biologist, Human Rezaei, who is specialized in prions, is involved in the project. His participation allows us to compare the model to results of in vitro experiments, but that is not all. This relationship is scientifically enriching for both of us. The mathematician's point of view leads the biologist to ask more questions and explore new leads. Conversely, the biologist submits unusual mathematical problems, revealing a new aspect of an equation that has already been studied in great depth. It is not always easy because we tend to reason according to the logic of our respective fields, but it is very stimulating. However, this requires a real investment, in terms of time, which is generally not recognised in each discipline. Accepting multi-disciplinary profiles is one of Inria’s greatest strengths. The fact that the ERC appreciates this interaction with the field of biology confers a form of recognition to this multi-disciplinary approach and a certain freedom in our research. That said, I haven't become a biologist at all and I still belong in applied mathematics!

"The challenge consists of modelling polymerisation using partial and aggregated data"

Applying partial differential equations to biology is a relatively recent approach because the problems are much more complex than in physics. Consequently it is very difficult to "match" a model to an experiment. It has to be streamlined so that it is simple enough to be studied and complex enough to capture the essence of biological behaviour. But that is not the only difficulty. In the case of amyloidosis, there are many measurements that are partial or aggregated, such as the measurement of total polymers over time. The challenge lies in developing models that can be confirmed or refuted based on this data. To achieve this, Marie Doumic-Jauffret uses partial differential equations (models of   growth/fragmentation/coagulation) and compares this method with statistical and probabilistic approaches – a confrontation rich in information and well-adapted to this type of problem – as well as inverse problem techniques that use measurements of a phenomenon to select which model it obeys. In the Skipper project, research is applied to proteins, but the methods developed should be universal enough to produce results in other fields.