A new level of science: artificial evolution and complex systems

Share this article

Pierre Collet, professor in Computer Science at Strasbourg University, is one of the key opinion speakers at the BioWin Day 2016. He works on artificial evolution, the human physiome and complex system sciences. He is multitalented and has a clear vision of how science can be accelerated on a global scale. BioVox had the honor to interview him in advance of his speech at BioWin Day.
A linguist, a pilot, a surgeon — a scientist?

Prof. Pierre Collet has followed quite an atypical career path. Collet first obtained an MA in English and an MSc in computer science, studying both subjects at the same time. Next, he did research internships in computer science while simultaneously becoming a pilot. The airline company Euroways asked him to become assistant manager, an offer he couldn’t resist. However, after a couple of years in that position, he decided research was his true calling. At that point in time, he met a surgeon who specialized in rhinosurgery, a very risky procedure: If you go even slightly too far up the frontal sinus, you could scrape the brain, make your patient blind by damaging the orbitary wall or even kill him by causing a hemorrhage inside the cavernous sinus. Collet developed a tool that allowed surgeons to precisely track their instruments. With funding from the French electricity company EDF, Collet completed a PhD on computerized surgery.  He was trained to become an operating assistant, and for three years he helped the surgeon perform all surgical interventions. His work appeared in twenty publications within a three-year span, but when he applied for a position as assistant professor in computer science at a French university, he was told that his publications did not qualify because they were in medical journals, not in computer science journals. Eventually, he was hired as a researcher at the École Polytechnique in France. After four years, he managed to get a research position as assistant professor at the University of the Littoral Opal Coast. Finally, in 2007, he was appointed full professor at Strasbourg, his current position.

The science on how to perform science

“The new science of complex systems is behind everything, hard and soft sciences.

Complex Systems is a new science that studies how different parts of a system induce the emerging collective behavior of a system, in relation to its environment. Complex systems can be used in hard as well as in soft sciences and are widely applicable. Some examples are social systems formed (in part) out of people, the brain formed out of neurons, supramolecular constructs formed out of molecules and atoms and the weather formed out of airflow. “Complex systems address all sciences in the same way,” says Collet. “Complex systems is a new science, born in the middle of the 20th century. It develops around the ‘4P Factory,’ i.e., the Participative, Predictive, Preventive and Personalized paradigm. This can, for example, be used to accelerate healthcare development. Picture the following: You have a cohort of 100,000 people for 20 years; in this group, 10 people will develop a rare disease. Based on this information, we will try to find out if there are health trajectories that are common to all 10 people who develop the rare disease. Maybe they have a common mutation on a gene? But you might have 5,000 people with the same gene mutation. Why do only 10 of them get the disease? We will compare their trajectory, which might show a series of similar events, e.g., they all got measles between the age of 4 and 6, and later caught the herpes virus. So, in the future, we can predict with a very high probability that someone with the said mutation, who got measles between the age of 4 and 6, might risk developing the rare disease if he catches herpes… Then we must try to find a way to modify the health trajectory so that the patient can escape this disease.”

People think that this is still science fiction, but at the BioWin Day I will give some impressive results. Predicting future disease occurrence is what medicine has been trying to do for a long, long time.

Connecting brains worldwide

The idea sounds plausible, but how can you make this a reality? Collet answers: “In collaboration with Paul Bourgine, a preeminent French scientist, we are setting up a worldwide Complex Systems Digital Campus (CS-DC). In 2012, we started to collect universities, and in 2013 we filed an application to UNESCO to create a UniTwin, a university twinning network. These UniTwins usually have around 3–5 universities working together on a specific domain. But, because the topic of complex systems is so wide, we got to UNESCO with a file of 73 universities, which was validated by Irina Bokova (Director-General of UNESCO) at the end of 2014. We have been adding roughly one more university every week, so we are at 130 universities right now. These universities have between 3–4 million students, and around 4,000 researchers joined the CS-DC UniTwin. We organize them in the Complex Systems Digital Campus, which enables teaching and performing research on complex systems on a worldwide level. It’s an e-campus. Everything is online and connects universities in Europe, South America, Africa. We are now also incorporating India, China and North America. The first world conference on the Complex Systems Digital Campus took place last year and was a great success.”

An embryo in a computer

“To bring personalized health to a higher level,” explains Collet, “biologists, bioinformatics and computer scientists are developing what we call a human physiome. This is a set of equations that can replay a human being inside a computer from the embryo to the death of the patient. By 2040, we should have the first version of a complete human physiome.”

“Everyone already knows the first equation that models embryogenesis: You start with the first cell, this one divides into two, then into four, eight, sixteen cells and so on. However, beyond a certain stage, the cells will differentiate. The first cells that develop in the egg are in touch with the egg yolk. But as the cells multiply, they will assemble in layers; the first layer will be in direct contact with the egg yolk, but the second layer will not have the same access. These slightly different conditions will be interpreted by the DNA to induce differentiation of the cells. As the embryo is developing, in order to match what we observe, we are changing our equations, more precisely named integro-differential equations, so that in the end we can replay the development of an embryo in a computer by using a sequence of equations. We develop the embryo in silico until organs start to appear in the embryo. The lumps of cells that will form a certain organ are then replaced by a partial model representing the organ. Many research teams have been working for many years on partial models of the liver, the kidneys, the heart and all the other organs. At this point, we get into a second stage where we are not only simulating an embryo, but an entire organism, a small baby, a human being. The idea is to go on to the adult stage. Physiomes attached to a newborn and subsequently modified to reflect a patient’s health trajectory can later be used to test how someone will react to a drug in order to develop personalized medication.”

Artificial evolution creates new drugs

Evolution is a very creative process. There are animals living under the ground, animals living in the air, animals living in water — animals or bacteria exploiting different ecological niches everywhere. Extremophile bacteria live in liquid asphalt lakes! Artificial evolution is a way to make computers creative, Collet says: “When you implement artificial evolution in a computer, the computer will start to render solutions that nobody else has ever thought about before. Ever since 2000, computers that are implementing artificial evolution are human-competitive in designing solutions, meaning that they are equal to or better than human engineers. This is now a very well established fact (cf. John Koza’s book Genetic Programming IV). We can use artificial evolution to invent new molecules, new drugs. In my team, we are developing simulators for immune systems, and as soon as we have an immune system in a computer, we can develop new drugs on this immune system using artificial evolution, or even develop an immune system through artificial evolution that reflects observed data.”

We need money to make it big

“The new science of complex systems is behind everything, hard and soft sciences,” Collet concludes. “It is a new science on which all recent scientific Nobel laureates and Fields medalists are working, but, for instance, at Strasbourg University, which hosts 4 Nobel Laureates who have been working on complex systems science, there’s no BSc, MSc or PhD in complex systems. We are trying to set this up with the UNESCO UniTwin program. However, we need industrial partners who want to finance this gigantic project. We need companies to invest in complex system sciences, because complex systems is now the new scientific paradigm that truly drives current and future research.”