An in silico experimental evolution platform

We are pleased to announce the third Aevol tutorial that will be held during the Artificial Life conference in Cancun.

In silico experimental evolution with the Aevol software


  • Guillaume Beslon
  • Carole Knibbe
  • David P. Parsons
  • Dusan Misevic

Description of the tutorial

Aevol is a digital genetics model in which populations of digital organisms undergo variation and selection, creating a Darwinian dynamics. By modifying the characteristics of selection (e.g. population size, type of environment, environmental variations) or genetic variation (e.g. mutation rates, chromosomal rearrangement rates, horizontal transfer), one can study experimentally the impact of these parameters on the evolved organisms. In particular, since Aevol integrates a realistic model of the genome, it allows for the study of various structural properties of the genome including gene number, syntheny, and proportion of coding sequences. The Aevol simulation platform also includes tools to analyze phylogenies and measure characteristics of the organisms and populations during their evolution.

Because of its non–trivial genotype–phenotype–fitness mapping, Aevol is particularly adapted to the study of second–order selection for genetic robustness, variability, or evolvability. Second–order selection generally refers to the selection for traits that do not provide a direct fitness benefit, or are even costly, but are beneficial in the long run. For example, genetic robustness does not increase the number of offspring that the individual produces, but it does affect the probability that these offspring will be as fit as their parents, leading to a higher chance of success of the whole lineage (survival of the flattest).

Developed in Lyon and Paris since 2005 (see below the reference list), Aevol is now a mature and stable platform that can be used by other teams to tackle their own questions. In this tutorial we will provide potential users of the software with all the necessary information to test the software, as well as design and start their own experiments. In the first part of the tutorial, we will present the structure of the software and the artificial chemistry used to compute a phenotype from a genotype. We will also insist on the «in silico experimental evolution» methodology and on the similarities with in vivo experimental evolution. The second part of the workshop will be a mini–lab during which the participants will have a chance to run either some classical evolutionary experiments or new ones of their own design. Finally, we will present the general overview of Aevol's code structure in order to help more advanced users to develop extensions of the initial model.

Tentative Schedule

  • General overview of the experimental methodology and of the Aevol software (30 min).
  • Details of Aevol usage: tools and parameters (30 min).
  • Labwork: experimenting with aevol (personal machine needed; Aevol is available on Linux and MacOS X platforms) (1h30).
  • Going further: a quick tour on Aevol programming (C/C++ skills required) (30 min).


  • A background in evolutionary biology or in artificial evolution is welcome but not mandatory. Programming abilities in C/C++ will be useful for the last part of the tutorial but not necessary for the other parts.
  • Participants are invited to bring a laptop computer. Even a netbook would do, but a medium‐range computer would allow attendants to accomplish more during the lab part of the tutorial. Both Linux and MacOS X are supported, but not Windows. The organizers can provide a few computers for the tutorial, please contact us beforehand if you would like to arrange one for you.

Brief curriculum of the organizers

Guillaume Beslon is a professor at the Computer Science Department of the Institut National des Sciences Appliquées de Lyon and the former head of the Rhône-Alpes Institute of Complex Systems. He is the leader of the INRIA Beagle Team that conducts researches on Computational Biology and Artificial Evolution. He started his research carrier by working on bio‐inspired computing (spiking neural networks, genetic algorithms) and now conducts projects on digital genetics (in particular the Aevol project) and the modeling of cellular processes (stochasticity of gene expression). He is the leader of the European EvoEvo project (2013–2016) that aims at developing new evolutionary paradigms by taking inspiration from microbial evolution ( He published more than 40 articles, mainly in international journals in biology and computational biology and in artificial life conferences (ALIFE, ECAL).

Carole Knibbe is an associate professor in the computer science department of Université Lyon 1 (France) and a computational biologist in the INRIA Beagle team. Her initial training is in bioinformatics and modeling of biological systems. During her PhD, she built the Aevol computational framework to study the structural evolution of microbial genomes. Her research interests are the evolution of genome size and structure, the evolution of gene networks, the evolution of mutational robustness and variability, and the evolution of evolvability. She builds computational and mathematical models of evolving populations to study these properties. She recently coordinated a two-years interdisciplinary project called «Analyze, simulate and experiment the evolution of bacterial genomes», involving five partners from microbiology, bioinformatics, mathematics and computer science.

David P. Parsons is a Research Engineer at the French National Institute for Research in Computational Sciences (INRIA) and a lecturer at the French National Institute for Applied Sciences (INSA) in Lyon, France. During his PhD in the INRIA Beagle team, he studied the mechanisms responsible for and the effects of second-order selection using Aevol. He also developed much of the current version of the software.

Dusan Misevic is a researcher in evolutionary biology at the French National Institute for Health and Medical Research (INSERM) and a lecturer at the Center for Research and Interdisciplinarity, both in Paris, France. He did his PhD on the evolution of sex, testing various hypotheses about the benefits of sex and examining the effects of recombination on genetic architecture in the classic in silico system, Avida. Currently he is managing the project financed by French National Research Agency (ANR) on the evolution of cooperation, the evolution of plasmid conjugation, and the interactions between the two. He has been working with and occasionally developing Aevol for the past 5 years. He has a strong background in experimental evolution and as of recently is also working with microbial systems. He has published well-received papers at both Alife and ECAL conferences, as well as in various biology journals.

Vincent Liard is a computer engineer at Inria, the French National Institute for Research in Computational Sciences. He is devoted to code maintenance and improvements. He also takes part into performance monitoring and optimization.