Research

Model the living structures:
Biology

Based on the conviction that living system modeling is one of the greatest scientific challenges of the 21st century, INRIA is putting its competence in information processing and modeling to work on this issue. Its research teams propose innovations in the extremely broad field of life sciences that contribute to the understanding of natural phenomena, progress in protection of the environment and a better quality of life.

The diversity and complexity of life:
A growing challenge for INRIA research

Human red cells infected with malaria - © Institut Pasteur

Researchers in genomics, population dynamics, and plant and ecosystem studies are faced with common challenges:

• Complexity: from small components-molecules or proteins-to large scale equilibria, there are multiple facets to life. The variability is extreme. Evolution is also constant.
• The difficulty or even impossibility of performing experiments. For example, it is impossible to trigger a biological catastrophe in order to assess its impact on an ecosystem. Computing and modeling can provide significant contributions in these fields.

This is why numerous INRIA teams are working in close partnership with biologists with the following goals:

• To manage and exploit massive, complex data. Research accumulates information and data, and access to this information and data must be simplified. This is the case for example in genomics where databases grow at the same time as sequencings are completed. Computers are thus necessary to sort the information, search inside data stocks, identify correlations, highlight links and so forth.
• To model phenomena in order to simulate, understand, analyze and validate the biologists' theories. COMORE has specialized in the study of complex ecosystems, by applying the automatic control approach to control fragile biological equilibria. The team is also participating in the GDYN ARC by bringing in its competence in dynamic modeling for the study of genic regulation networks, in collaboration with other teams. CONGE models large scale pandemics (malaria, tuberculosis, etc.). Models have already proven themselves in the field of public health for identifying the most efficient therapies. MERE puts applied mathematics at the service of water, an increasingly monitored resource.
• To simulate models in virtual reality in order to contribute to a better understanding of real phenomena. EVASION develops algorithms that produce 3D animations based on the observation of living beings. The team also participates in the Docking ARC, which resulted in particular in offering research scientists an interactive protein manipulation tool to test whether proteins can be coupled together.

Thus, more than ever, mathematics and computing are special partners for biology.

How to preserve our ecosystems

Ecosystems are fragile. Can modeling contribute to preserving them by identifying their reactions to external stimuli?

Control and regulation are key issues in the management of renewable living resources. Human activities have an impact on biological systems-whether marine plankton or bacterial populations-that could be irreversible if it was not controlled. This is why COMORE research scientists had the idea of applying the automatic control approach based on control and regulation, to model the way such systems evolve, identify their adaptation capabilities and understand them better.

This approach results in the development of models, the complexity of which is equally linked to the number of parameters involved and to their evolution in time. Indeed, dynamics is one of the key scientific issues of this work.

One of the major themes of COMORE research is phytoplankton development. In this field, observations at sea are complex. Researchers thus devised equipment that can be used to test and validate their models, in collaboration with a CNRS team located at Villefranche sur Mer. The phytoplankton growth environment is reproduced in the laboratory within a bioreactor, a chemostat-a real dynamic, living system.

Chemostat

Strict control of the input variables, such as the temperature, that is impossible in a natural environment, makes it possible to perform experiments and test the validity of modeled phenomena. This step is crucial. Models can only be used by biologists if they translate reality precisely. Using the chemostat, COMORE research scientists are improving the efficiency of their phytoplankton growth models that will make it possible to better understand marine life and respect its equilibria.

COMORE research on the dynamical modeling of biological systems and their control had a great many other applications: remote monitoring of anaerobic fermentation processing of food processing waste, in the framework of the Telemac European network, fishing management with IFREMER and biological control (of aphids for example) with INRA. Today, COMORE is also interested in the dynamics of a very specific population-genes-in the framework of the Gdyn ARC.

Understanding epidemics better

What can modeling and automatic control contribute to the fight against the main epidemic plagues such as tuberculosis and malaria, that still ravage certain parts of the world?

Malaria still kills two million people every year. For a long time, experts have studied its transmission, looking for a way to eradicate it. However, reality forced them to reorient their work. The question is how this genuine biological system, with its mosquito, parasite and sick patient populations, managed to resist all attempts to fight it.

Moustique Anopheles gambiae
© Institut Pasteur

The CONGE team has special links with sub-Saharan Africa. It has long been interested in these questions and develops malaria models based on statistical data and doctors' hypotheses on the mosquito populations, the concentration of parasites in an organism, the reaction of the immune system and so on.

The contributions of this work are essential. The description of phenomena and identification of significant parameters, which constitute the first steps in modeling, help us to better understand the epidemic. By applying automatic control methods, the team also tries to identify what controls and locks the system. How are adaptation and resistance reactions triggered? The team is trying to identify the weak points of the disease, which would make it possible to better assess the necessary public health initiatives. Should mosquito nets impregnated with insecticide be used, for example? Do they not just reinforce parasite resistance while lowering human resistance? How does immunity develop? Why is the disease more severe in certain case, for example in the often fatal neuro-malaria form?

Plasmodium falciparum - © Institut Pasteur

The CONGE team works on biological data collected in situ. However, certain indicators that seem to be essential in the development of the epidemic are not available. For example, the probability of developing malaria in case of mosquito bite is not known. To assign a value to such variables that cannot be observed, researchers use the "observer theory" and develop algorithms that make it possible to assign such a value via computations.

CONGE research scientists work in collaboration with the Tropical Medicine Institute of the Army Health Service, the Pasteur Institute, the Lyon Veterinary School, and the Institute for Research for Development in Montpellier. They apply their research to other epidemics: especially tuberculosis by studying transmission structures, the Ebola virus based on data on three epidemics in Congo, etc.

If biology is not the sole work field of CONGE, it is assuming an increasing importance.

Improve the management of our natural resources

Can scientific computing and modeling help better manage water, a vital resource for the planet?

Water consumption is steadily growing in France and in other developed countries. This is especially true of industry and food-processing consumptions, which is often the cause of high concentration of such pollutants as nitrates and phosphates.

Project MERE was founded in 2004 to work on water problems in general. The project's first works concern biological cleanup processes. In these different techniques-activated sludge, anaerobic digestion, and so on-an ecosystem consisting of bacterial populations is used to digest water pollutants. Wastewater is poured into a vat where the pollutants and biomass concentrate, which triggers successive biological reactions that result in the digestion of pollutants. Such living systems are complex, sometimes unpredictable and very unstable. Biological reaction modeling and the application of control techniques are very useful to improve their reliability.

One of the original aspects of MERE research is linked to the engineering of such systems. Shouldn't two vats connected in series be more efficient than a single vat? Do the size and configuration of the vat (cylindrical, cubic, etc.) have an impact on the efficiency of the bio-digestion? Applying classical engineering techniques to these questions clashes with the living realm. For example, in the case of a vat used for a chemical reaction, the mixture will be made homogeneous by giving it an energetic and regular stir. This is not possible for a bio-reaction, since too much movement could stop the activity of bacteria. Adapting technology to living beings requires, in this case as in others, the development of specific mathematical tools and modeling methods. The MERE team is especially working on the spatial analysis of the cleanup ecosystems reactions and their efficiency in different spots of the vat. This makes it possible to better define an optimal shape for the vat.

MERE is the first INRIA/INRA joint research project. MERE research scientists thus have access to the experimental equipment of INRA and can validate their work. This equipment also offers an exceptional opportunity to analyze ecosystems under laboratory conditions. The objective of the team is to obtain processes that are at the same time efficient, easily implemented and economically viable. In effect, the social aspect is increasingly important when European regulations are imposing more and more stringent restrictions on wastewater disposal in the environment.

MERE research scientists are also working in close partnership with teams of African mathematicians who are developing modeling tools for other water-related problems. In this framework, MERE welcomes African doctoral candidates and post-docs.

Can virtual reality contribute to progress in biology?

Computer graphics is not just for cartoons and videogames. The EVASION team specializes in the simulation of animated natural objects for virtual reality and 3D computer graphics. The team has multiple contributions to the life sciences. The visualization of hard to observe phenomena is one of the obvious applications of this work. For example, plant growth can be accelerated and we can witness the meeting of two molecules. Beyond that, certain applications are more surprising. Based on animal movies, researchers developed a model that can be used to analyze and reproduce the movements of animals in computer graphics. This is now of interest to specialists. In effect, it is not easy to study the functioning of muscles in a wild animal. Classical analysis techniques are inoperative: the animal would have to be equipped with sensors, which is obviously very difficult, if not impossible, in nature. As soon as some distance must be kept, observation and films are the only available sources of information. Owing to EVASION work, movements are analyzed and modeled based on such images shot in nature. From a 25 images per second, 2 hour movie, are extracted four or five key images that will be used to reconstruct a fluid movement.

3D models of a cheetah

The tool thus obtained makes it possible to better understand the physical and biomechanical constraints of animal movement. It is currently of interest to an animal park.

EVASION also works on the perception of facial expressions in collaboration with the Neurology Clinic of the teaching hospital of Geneva, the Psychology Department of the University of Geneva and the Neurophysiology Department of the Catholic University of Louvain. The objective of medical doctors is to identify the areas of the cortex involved in the perception of facial expression as a function of the viewing direction. This faculty could be one of the first to become impaired in case of neurodegenerative disease. The contribution of EVASION was to develop a software tool that makes it possible to realistically give faces predefined viewing directions, based on a series of face images used during clinical trials. Doctors can thus present patients with the precisely needed visual stimulus.

A Layered Model of a Virtual Human Intestine for Surgery Simulation

In addition to images, EVASION develops genuine virtual worlds-3D universes with interaction. For example in surgical training applications, the user holds controls and sees a virtual organ deform instantly as a result of the user's gestures.

The scientific challenges of this work are multiple. A level of simplification sufficient to translate and generate the complexity of living beings must be identified. Moreover, for the simulation to be meaningful, visual rendering must be realistic. EVASION is thus working on appearance. Real time interaction is a third challenge for the team by requiring the development of deformable models close to those used in engineering but thousands of times faster, thus opening up wide possibilities for medical applications of virtual reality.

Visualize the infinitesimal

How can we know if two molecules can pair up?

The biological world makes abundant use of "keys" and "locks" consisting of large, skillfully folded molecules. Intercellular communication is for instance effected through cell membranes by the meeting of proteins, one coming from the outside of the cell and the other from the inside. If they correspond to each other, the two proteins pair up-this phenomenon is called docking. They then change shapes and properties, which triggers an action.

Given the scale, docking is extremely difficult to observe. We cannot see what happens at the interface between the molecules at the contact points between atoms. Moreover, many molecules are only known via genome decryption. Their role is unknown and scientists are trying to find out precisely which lock they unlock.

Optimization of protein-protein docking via high performance computing and visualization

Simulation is an answer to this problem. We could make a computer systematically try out all the positions in order to identify the possible meetings. This represent a colossal amount of computation that is out of reach of the most powerful machines. The solution proposed by the Docking ARC, in which the Geometrica, EVASION and ISA teams participate in partnership with the "Membrane assembly dynamics" teams of CNRS, consists in filtering the candidate molecules through a cascade of screens to only keep the couples that have an actual chance of pairing up, due to their shape. A precise simulation can then be run on such couples.

This work has already resulted in software that lets biologists visually interact in 3D and real time with virtual molecules, by simulating the contacts and deformations when molecules get close to each other. With this research, biologists will thus progress faster in understanding the mechanisms at work in cells and living organisms.

Anticipate the reactions of organisms

Can population dynamics applied to genomics make it possible to better understand the evolution of living beings?

Biologists have shown that the properties of an organism and its evolution depended on interactions that occur at different levels of gene and protein expression. The state of certain genes can inhibit or activate other genes. Scientists call this genic regulation networks. This discovery was done on bacteria which present the advantage of growing and renewing very rapidly. For example, under certain unfavorable circumstances, Bacillus subtilis, a model bacteria for the study of other pathogenic bacteria such as staphylococci and streptococci, stops growing, forms resilient spores and appears to settle in a survival situation while waiting for better days. Genic regulation networks seem to be responsible for this "decision" that depends on the environment, just like in ecosystems a regulation is turned on as soon as external conditions change.

Genetic Network Analyser

This is why research scientists had the idea of applying automatic control tools to better understand this phenomenon as well as other stress responses in bacteria. To begin this study, several INRIA teams pooled their competence within the GDYN ARC: COMORE specializes in dynamical population modeling, Helix is a bioinformatics project and Sosso an automatic control project. These teams are working in partnership with biologists of the Microorganism Adaptation and Pathogeny of the Joseph Fourier University in Grenoble and CNRS, of the Photosynthetic Organism and Environment laboratory of the ENS and CNRS, as well as with mathematicians of the Mathematics and Applications Department of the Haute Alsace University in Mulhouse.

This work resulted in the design of a qualitative simulation method for regulation network that was implemented in the GNA (Genetic Network Analyzer) software tool.