From Molecular Basis to Predictability and Control of Evolution

FB52 (Nordita, Stockholm)


Nordita, Stockholm

Armita Nourmouhammad (MPI for Dynamics and Self-organization) , Fernanda Pinheiro (University of Cologne) , Marta Luksza (Icahn School of Medicine, Mount Sinai)


Nordita, Stockholm, Sweden


Growing amount of molecular biological data combined with current advances in modeling of complex systems provide unprecedented opportunities to understand biological evolution in a quantitative way. A quantitative description of an evolving system is the first step towards prediction and control, and it opens new exciting directions for highly interdisciplinary research. The central questions are: (i) to what degree we can predict the outcome of biological evolution? (ii) what features of the system are predictable? (iii) which features confer predictive value to ultimately control an evolving population? This program brings together theoretical and experimental physicists, experimental biologists with an interest in quantitative modeling and mathematicians with interest in biological systems. We aim to create a dialog between researchers of different fields and to inspire future collaborations. In addition, further developments in this field would have significant translational impacts, e.g., by optimizing vaccines against evolving viruses, designing strategies for personalized cancer therapy and by providing insights to the problem of antibiotic resistance.

Speakers Include

  • Dan Andersson (Uppsala University)
  • Erik Aurell (KTH, Stockholm)
  • Claudia Bank (Instituto Gulbenkian)
  • Anne-Florence Bitbol (CNRS)
  • Simona Cocco (ENS and CNRS)
  • Jeff Gore (MIT)
  • Sidhartha Goyal (University of Toronto)
  • Oskar Hallatschek (UC Berkeley)
  • Chris Illingworth (University of Cambridge)
  • Mehran Kardar (MIT)
  • Joachim Krug (University of Cologne)
  • Michael Lässig (University of Cologne)
  • Peter Lind (Umea University)
  • Berenike Maier (University of Cologne)
  • Enzo Marinari (Sapienza University of Rome)
  • Matteo Marsili (ICTP)
  • Rémi Monasson (ENS and CNRS)
  • Thierry Mora (ENS and CNRS)
  • Alexandre Morozov (Rutgers)
  • Richard Neher (University of Basel)
  • Jakub Otwinowski (MPI Goettingen)
  • Olivier Rivoire (CNRS)
  • Mikhail Tikhonov (UW St. Louis)
  • Aleksandra Walczak (CNRS)
  • Martin Weigt (Sorbonne University)


If you want to apply for participation in the program, please fill in the application form. You will be informed by the organizers shortly after the application deadline whether your application has been approved. Due to space restrictions, the total number of participants is strictly limited. (Invited speakers are of course automatically approved, but need to register anyway.)

Primary application deadline: 28 Feb 2019

Applications accepted until the program is filled.

A minimum stay of one working week is required and we encourage participants to stay for a period of at least two weeks.

There is no registration fee.


Nordita can provide a limited number of rooms and apartments in the Stockholm apartment hotel BizApartments. If you are interested in this accommodation, please indicate when registering for the conference.

Travel Reimbursement

PhD students and young Postdoc fellows are eligible for travel grants to participate in the program. If you are interested in such a grant, please mark the corresponding field in the application form, briefly summarize your interest in the program in the comments field, and indicate an estimation of your expected travel expenses. Since only a limited number of grants is available, decision concerning the grants will be made on a case-by-case basis and you will be notified shortly after the application deadline.

More information about the schedule will be available here later.

Sponsored by:


CRC logo
    • 1
    • 2
      Speaker: Dan Andersson
    • 3
      Speaker: Michael Laessig
    • 4
      Impact of a periodic presence of antimicrobial on resistance evolution in a microbial population
      Antimicrobial treatments select for resistance, and once resistance becomes widespread, antimicrobials become useless. Resistance evolution within a host can be strongly affected by variations of antimicrobial concentration that may exist e.g. during a treatment. We are investigating these effects using stochastic models. First, in a microbial population of fixed size, we showed that fast alternations of phases with and without antimicrobial strongly accelerate the evolution of resistance, especially for large populations. Next, we considered microbial populations of variable size, in which we studied the impact of biocidal drugs, that kill microorganisms, and of biostatic drugs, that prevent microorganisms from growing. In both cases, we showed that the probability of treatment success, i.e. of extinction of the microbial population, strongly depends on the period of the alternations of drug absence and presence. Moreover, we showed that biocidal antimicrobials promote resistance more than biostatic ones. Finally, we found a population size-dependent critical drug concentration below which antimicrobials cannot eradicate microbial populations.
      Speaker: Anne-Florence Bitbol
    • 5
      How adaptive immunity governs co-evolution in microbes
      Features of the CRISPR-Cas system, in which bacteria integrate small segments of phage genome (spacers) into their DNA to neutralize future attacks, suggest that its effect is not limited to individual bacteria but may control the fate and structure of whole populations [1]. In our model, we find that early dynamics of a recently acquired spacer is largely independent of phage dynamics but crucially depends on the burst-size of phage infections. In contrast, the fates of high abundant spacers are strongly influenced by the feedback from phages that creates a time-dependent fitness landscape for the spacers. Taken together, we quantify the role of population parameters in maintaining phage and bacterial diversity where CRISPR-cas is in the play. [1]
      Speaker: Sidhartha Goyal
    • 6
      A statistical-inference approach to access an ‘interaction’ pattern among migrating cells
      In this talk I will discuss the development and implementation of a statistical-inference tool to investigate extracellular and contact independent cell-to-cell communication, that possibly yield to collective migration. This tool, based on the maximum-entropy principle, is entirely statistically-driven rather than biologically-driven: it does not rely on an a priori knowledge of the biochemical interactions among the migrating cells. Instead, we leverage the statistical information stored in empirical observations on cell motility, e.g., quantitative analyses of cell speed and directionality obtained in control environments, such as microfluidic devices. I will finally present some examples of application, where we focus on immune cells moving on a lab-on-chip under different conditions. References E. Agliari, E. Biselli, A. De Ninno, G. Schiavoni, L. Gabriele, et al., “Cancer-driven dynamics of immune cells in a microfluidic environment”, Scientific Reports, 4, 6639 (2014). E. Biselli, E. Agliari, A. Barra, F.R. Bertani, A. Gerardino, et al., “Organs on chio approach: a tool to evaluate cancer-immune cells interactions”, Scientific Reports, 7, 12737 (2017). E. Agliari, P.J. Sáez, A. Barra, M. Piel, P. Vargas, M. Castellana, “A maximum-entropy approach for leukocyte migration in lab-on-chip experiments”, to be submitted.
      Speaker: Elena Agliari
    • 7
      An uncertainty law for microbial evolution
      Higher order epistasis is of practical importance for antibiotic and HIV drug resistance. For instance, a triple mutant may lead to treatment failure, whereas the corresponding single and double mutant are harmless. The relative importance of pairwise versus higher order interactions have been subject to some debate. However recent empirical studies suggest that the latter type is important. We explored the relation between higher order epistasis and predictability by considering fitness graphs, i.e., directed hyper-cube graphs. For each one of the 193,270,310 four-cube graphs, we counted the peaks and checked if the graph implied epistasis by using our characterization of graphs with this property. We found a strong association between higher order epistasis and the number of peaks based on exact computation. The relevance for larger system, as well as for antimicrobial drug resistance, will be discussed.
      Speaker: Kristina Krona
    • 8
      Forecasting experimental evolution in Pseudomonas
      Previous work on the adaptive “wrinkly spreader” (WS) type in Pseudomonas fluorescens allowed us to develop models and rules to predict evolutionary outcomes. Knowledge of the genotype-to- phenotype map of the WS phenotype and of alternative phenotypic solutions to the problem of air-liquid interface colonization allows forecasting for related species. Equipped with the genome sequences of P. protegens and P. syringae combined with knowledge from work in P. fluorescens, predictions about the outcome of experimental evolution were made. Phenotypic predictions were successful in terms of the exopolymeric substances (EPSs) structural components used to colonize the air-liquid interface caused by mutational activation of diaguanylate cyclases (DGCs). As predicted, the most common mutation type was loss-of-function mutation followed by less common promoter mutations and promoter capture events. Although up to 39 different DGCs are encoded in these genomes the relative rates of the top two or four was successfully predicted. Individual mutations could not be predicted, but the mutated regions were conserved between species with the majority of mutations expected to disrupt intermolecular or interdomain interactions. Mutational hot spots were not conserved between species. This work shows the potential of extending one of the best- characterized experimental evolution systems to other species allowing true testing of evolutionary forecasts. Frequent horizontal gene transfer and gene loss of DGCs and genes encoding structural EPSs creates a large diversity among different Pseudomonas species meaning that the prediction is different for each species. This allows an iterative workflow of prediction, experimental evolution and model improvement to explore the limits of evolutionary forecasting. There is potential for collaborations on many levels including development and testing of methods for predictions on different biological levels, theoretical considerations and modelling as well as comparisons with other experimental systems.
      Speaker: Peter Lind
    • 9
      Repeatability and predictability in microbial evolution
      RNA viruses like HIV of influenza virus evolve rapidly and thereby evade human immunity. While molecular evolution proceeds in the high dimensional space of possible genomes, independent realization of the viral evolution frequently proceed via similar mutational patterns and predictably revert to ancestral states once selection pressures subside. I will discuss the extent and limit of such repeatability using longitudinal deep sequencing data of HIV populations. Related patterns can be used to predict which influenza virus variants are most likely to succeed and circulate in future seasons. In bacteria repeatability to is typically limited to very strong pressure such as antibiotic selection and is often found at the level of genes and pathways rather than individual positions in the genome.
      Speaker: Richard Neher
    • 10
      Population size and the predictability of antibiotic resistance evolution
      Although evolution is inherently stochastic due to chance events, such as mutations, it is not fully random due to deterministic consequences of natural selection and constraints imposed by the fitness landscape. A better understanding of the causes and constraints of evolution will help to control the evolution of unwanted phenotypes, such as antibiotic-resistant pathogens. I will present results from evolution experiments with the notorious antibiotic resistance enzyme, TEM-1 beta lactamase, in the presence of a novel antibiotic. We use both in vitro (only TEM) and in vivo (TEM + host) evolution experiments to address the effect of population size on the tempo and mode of adaptation. The results from in vitro experiments show greater adaptive heterogeneity and surprisingly higher resistance in small than in large populations, reflecting the rugged fitness landscape of the enzyme. In the in vivo experiments, we find no adaptive benefit for small populations – suggesting a more smooth adaptive landscape, but differential effects from population size on the repeatability of different mutation classes: large populations show more parallel SNPs, while small populations show more parallel large genomic deletions and duplications. We are testing the hypothesis that this divergent pattern of parallel evolution derives from clonal interference benefitting lower- rate, but larger-benefit SNPs in populations large enough for both classes of mutations to occur.
      Speaker: Arjan de Visser
    • 11
      Speaker: Fernanda Pinheiro
    • 12
      Toy models of evolved adapting machines
      The internal state of living systems, from single proteins to whole cells and organisms, needs to respond to the state of their environment. At odds with physical systems, the Hamiltonian of living systems is shaped by evolution in order to make this interaction optimal. Within a highly stylized model, I shall describe systems whose internal state is maximally informative about the state of the environment, and compare them to normal physical systems. (see
      Speaker: Matteo Marsili
    • 13
      Some recent progresses in Neural Networks & Machine Learning via Statistical Mechanics
      In this talk, keeping a statistical mechanical perspective, I will first revise the link between learning in Restricted Boltzmann Machines and retrieval in Hopfield Neural Networks. Then, I will focus on the importance in the choice of the nature of both neurons and synapses (e.g. digital vs analog) for both learning and retrieval mechanisms. Finally, I will discuss some variants of the Hebbian paradigm that allow the network to saturate the critical storage capacity, reaching Gardner's bound for symmetric networks. Minimal Reference: [1] Barra, A., Genovese, G., Sollich, P., & Tantari, D. (2018). Phase diagram of restricted Boltzmann machines and generalized Hopfield networks with arbitrary priors. Physical Review E, 97(2), 022310. [2] Fachechi, A., Agliari, E., & Barra, A. (2019). Dreaming neural networks: forgetting spurious memories and reinforcing pure ones. Neural Networks. [3] Agliari, E., Barra, A., Galluzzi, A., Guerra, F., & Moauro, F. (2012). Multitasking associative networks. Physical review letters, 109(26), 268101.
      Speaker: Adriano Barra
    • 14
      Poster Flash Talks W1
    • 15
      Poster Session W1
    • 16
      Antigenic evolution of viruses in host populations
      To escape immune recognition in previously infected hosts, viruses evolve genetically in immunologically important regions. The host’s immune system responds by generating new memory cells recognizing the mutated viral strains. Despite recent advances in data collection and analysis, it remains conceptually unclear how epidemiology, immune response, and evolutionary factors interact to produce the observed speed of evolution and the incidence of infection. Here we establish a general and simple relationship between long-term cross- immunity, genetic diversity, speed of evolution, and incidence. We develop an analytic method fusing the standard epidemiological susceptible-infected-recovered approach and the modern virus evolution theory. The model includes the factors of strain selection due to immune memory cells, random genetic drift, and clonal interference effects. We predict that the distribution of recovered individuals in memory serotypes creates a moving fitness landscape for the circulating strains which drives antigenic escape. The fitness slope (effective selection coefficient) is proportional to the reproductive number in the absence of immunity R0 and inversely proportional to the cross-immunity distance a, defined as the genetic distance of a virus strain from a previously infecting strain conferring 50% decrease in infection probability. Analysis predicts that the evolution rate increases linearly with the fitness slope and logarithmically with the genomic mutation rate and the host population size. Fitting our analytic model to data obtained for influenza A H3N2 and H1N1, we predict the annual infection incidence within a previously estimated range, (4-7)%, and the antigenic mutation rate of Ub = (5 − 8) 10−4 per transmission event per genome. Our prediction of the cross- immunity distance of a = (14 − 15) aminoacid substitutions agrees with independent data for equine influenza. Importantly, we demonstrate that variation in basic reproduction ratio R0 is mostly responsible for the variation among strains, and explain the observed inverse correlation between the substitution rate and the time to the most recent common ancestor.
      Speaker: Igor Rouzine
    • 17
      Implications of HA protein stability on influenza A virus evolution
      Thanks to their rapid evolution, influenza A viruses are able to seasonally infect a significant fraction of the human population. This evolution is driven by strong immune selection in partially immunized hosts, coupled with an error-prone RNA polymerase. The hemagglutinin (HA), a key viral surface protein, is the major target of host antibodies and thus the site of most variation. How do functional constraints of the HA, such as protein stability, shape evolutionary and transmission dynamics? In this work, we formulate a cross-scale mathematical model that considers molecular properties of the HA, within-host evolutionary dynamics, and between-host transmission. We characterize the emerging properties of flu, and glean important insight into its long-term evolution.
      Speaker: Roy Chadi
    • 18
      Welcome W2
    • 19
      Gene transfer across bacterial subspecies rapidly navigates a complex fitness landscape
      Horizontal gene transfer is an important factor in bacterial evolution that can act across species boundaries. Yet, we know little about rate and genomic targets of cross-species gene transfer, and on its physiological and selective effects in the recipient organism. Here we address these questions in a parallel evolution experiment with two Bacillus subtilis subspecies of 6.8% sequence divergence. We observe rapid evolution of hybrids by lateral gene transfer, and we show that these dynamics involve physiological and evolutionary adaptation. The genome-wide uptake of orthologous genes, together with insertions and deletions of accessory genes and de-novo mutations, generates genome evolution. Each recipient population replaces about 12% of its core genes, and 51% of core genes are replaced in at least one population. While evolved hybrids show a net loss of gene expression compared to the ancestral recipient population, we identify a set of genes whose upregulation is predictive of hybrid fitness. Moreover, the co-occurrence statistics of orthologous transfers reveals a broad network of fitness epistasis between essential genes. Together, these results show that gene transfer can bridge epistatic barriers between closely related species along multiple high-fitness paths.
      Speaker: Berenike Maier
    • 20
      Mutational robustness as an evolutionary benefit of recombination
      Mutational robustness quantifies the effect of random mutations on fitness. From the point of view of fitness landscapes, robust genotypes form neutral networks of equal fitness. Using deterministic population models it has been shown that selection favors genotypes inside such networks, which results in increased mutational robustness. Here we demonstrate that this effect is massively enhanced by recombination. Our results are based on a detailed analysis of mesa- shaped fitness landscapes and numerical simulations on percolation landscapes and an empirical landscape. We argue that the favorable effect of recombination on mutational robustness is a highly universal feature that may have played an important role in the emergence and maintenance of mechanisms of genetic exchange.
      Speaker: Joachim Krug
    • 21
      Predicting the fitness from protein sequence data
      I will present recent works on fitness predictions from Inferred Potts parwise models, bechmarking then on artificial data obtained from simple protein models and comparing withe experiments.
      Speaker: Simona Cocco
    • 22
      E. coli from fermentation to respiration: Pareto helps in a difficult choice
      Living cells react to changes in growth conditions by re-shaping their proteome. This accounts for different stress-response strategies, both specific (i.e., aimed at increasing the availability of stress mitigating proteins) and systemic (such as large-scale changes in the use of metabolic pathways aimed at a more efficient exploitation of resources). Proteome re-allocation can, however, imply significant bio- synthetic costs. Whether and how such costs impact the growth performance are largely open problems. Focusing on carbon-limited E. coli growth, we integrate genome-scale modeling and proteomic data to address these questions at a quantitative level. It turns out that optimal growth results from the tradeoff between yield maximization and protein burden minimization. Empirical data confirm that E. coli growth is close to Pareto-optimal over a broad range of growth rates. Our findings provide a quantitative perspective on carbon overflow, the origin of growth laws and the multidimensional optimality of E. coli metabolism.
      Speaker: Enzo Marinari
    • 23
      Statistical genetics and DCA on the (bacterial) genome scale
      Speaker: Erik Aurell
    • 24
      Equivalent exposure histories produce wide spectrum of multidrug resistance outcomes contingent on drug order
      Historical contingency has been shown to affect drug resistance evolutionary trajectories. However, the extent of this effect has been unclear. Here we address this by exposing E. coli to 24 equivalent drug histories, consisting of one drug-free and three drug-containing epochs, only altering the order of exposure. We further repeat the experiment by introducing the T4 bacteriophage, since bacteriophages are increasingly being used to treat bacterial infections alongside antibiotics. We find dramatic differences between histories in resistance evolutionary patterns, with outcomes ranging from virtually no resistance to virtually complete multidrug and phage resistance. We also identify factors responsible for desirable and poor outcomes. For instance, the presence of bacteriophage almost invariably results in a strong reduction in antibiotic resistance. These results show that historical contingency directs drug resistance trajectories to highly variable outcomes with predictable patterns that could be exploited in therapy optimization
      Speaker: Johannes Cairns
    • 25
      Speaker: Marta Luksza
    • 26
      Stochasticity and predictability in within-host viral evolution.
      Evolutionary processes in infectious disease occur on a variety of scales, from that of the single cell to that of global populations. At the level of within-host infection, absolute viral populations are typically large, such that selection acting on a population might be expected to lead to repeatable and deterministic patterns of evolution. We describe three exceptions to this behaviour. Firstly, examining the evolution of the influenza virus, we describe a case of chaotic short-term evolution, where small differences in the initial state of the population lead to substantially different pathways to adaptation. Secondly, considering data describing the within-host evolution of HIV, we consider the extent to which the fitness landscape is host-specific, such that distinct patterns of evolution are observed within each individual. Finally, we present work on the consequences of mutational load for viral adaptation, showing how variation in the genetic background upon which a beneficial mutation arises has both systematic and stochastic effects upon its fate.
      Speaker: Chris Illingworth
    • 27
      Protein sequence landscapes: from data-driven models to protein design
      n the course of evolution, proteins undergo substantial changes in their amino-acid sequences, while conserving their three-dimensional fold and their biological functionality. Modern sequencing techniques provide us with increasingly large families of evolutionary related proteins. Such data can be used to infer statistical models of sequence variability. I will discuss the surprising efficiency of models including pairwise epistasis (Potts models / Markov random fields), which are able to reproduce non-fitted statistical features of protein families, and therefore open ways to evolution-guided approaches to protein design.
      Speaker: Martin Weigt
    • 28
      Adaptation of olfactory receptor abundances to environmental changes
      Olfactory receptor usage is highly heterogeneous, with some receptor types being orders of magnitude more abundant than others. Experimentally, increased exposure to odorants leads variously, but reproducibly, to increased, decreased, or unchanged abundances of different activated receptors. We demonstrate that this diversity of effects can be understadood under the principle of efficient coding of olfactory information when sensors are broadly correlated, and provide an algorithm for predicting which olfactory receptors should increase or decrease in abundance following specific environmental changes. Finally, we give simple dynamical rules for neural birth and death processes that might underlie this adaptation.
      Speaker: Remi Monasson
    • 29
      Towards infering Potts models for evolutionary correlated sequence data
      Global coevolutionary models of protein families have become increasingly popular due to their capacity to predict residue-residue contacts from sequence information, but also to predict fitness effects of amino-acid substitutions or to infer protein-protein interactions. The central idea in these models is to construct a probability distribution, a Potts model, that reproduces single and pairwise frequencies of amino acids found in natural sequences of the protein family. This approach treats sequences from the family as independent samples, completely ignoring phylogenetic relations between them. This simplification is known to lead to potentially biased estimates of the parameters of the model, decreasing their biological relevance. Current workarounds for this problem, such as re-weighting sequences, are poorly understood and not principled. Here, we propose an inference scheme that takes the phylogeny of a protein family into account in order to correct biases in estimating the frequencies of amino-acids. Using artificial data, we show that a Potts model inferred using these corrected frequencies performs better in predicting contacts and fitness effect of mutations.
      Speaker: Pierre Barrat-Charlaix
    • 30
      Inferring effects of mutation on fitness: testing for context dependence and invariance of mutation effects
      I will present current research on developing methods for inferring and predicting the distribution of mutation effects and how population genomics surveys can be utilized to infer the distribution of mutation effects on fitness.
      Speaker: Thomas Bataillon
    • 31
      Poster Flash talks W2
    • 32
      Poster session W2
    • 33
      A universal biophysical model of codon bias
      Frequencies of synonymous codons (nucleotide triplets in the gene which are translated into amino acids by the ribosome) are typically non-uniform, despite the fact that such codons correspond to the same amino acid in the genetic code. This phenomenon, known as codon bias, is broadly believed to be due to a combination of factors including genetic drift, mutational effects, and selection for speed and accuracy of codon translation; however, quantitative modeling of codon bias has been elusive. I will present a biophysical model which explains genome-wide codon frequencies observed across 20 organisms. Our model implements detailed codon-level treatment of mutations, and includes two contributions to codon fitness which describe codon translation speed and accuracy. We find that the observed patterns of genome-wide codon usage are consistent with a strong selective penalty for mistranslated amino acids, while the dependence of codon fitness on translation speed is much weaker on average. Treating the translation process explicitly in the context of a finite ribosomal pool has allowed us to estimate mutation rates directly from the gene sequences. Overall, our approach offers a unified biophysical and population genetics framework for understanding codon bias.
      Speaker: Alexandre Morozov
    • 34
      Analysis of Evolutionary Constraints and Plasticity by Microbial Laboratory Evolution and Computational Models
      Biological systems change their state to evolve and adapt to changes in environmental conditions. Despite the recognized importance of characterizing the biological capacity to adapt and evolve, studies on biological evolvability and plasticity have remained at a qualitative level. To unveil how the course of evolution is constrained in high- dimensional phenotype and genotype spaces, we performed laboratory evolution under various (more than 100) stress environments, and changes in phenotypes and genome sequence were analyzed [1,2]. The results of these comprehensive analyses demonstrated that the expression changes were restricted to low- dimensional dynamics, while diverse genomic changes can contribute to similar phenotypic changes. Furthermore, to analyze the nature of evolutionary constraint, we performed computer simulations of adaptive evolution using simple cell models. Again, we found that cellular state changes in adaptation and evolution are generally restricted to low-dimensional dynamics. In this simulated dynamics of adaptive evolution, logarithmic changes in expression are shown to be proportional across all genes, with the proportionality coefficient given by the change in the growth rate of the cell, which was consistent with the experimental data [3,4]. Based on these results, we will discuss the nature of phenotypic plasticity and constraint in bacterial evolution, and possible strategies to predict and control the evolutionary dynamics.
      Speaker: Chikara Furusawa
    • 35
      Welcome W3
    • 36
      Speaker: Michael Laessig
    • 37
      Speaker: Benjamin Good
    • 38
      Path integral control theory for the control of populations
      Fisher et al PNAS 2014 establish a theoretical framework for the adaptive control of evolving populations. In particular, they connect the idea of adaptive therapy to the paradigm of stochastic optimal control. A stochastic treatment is necessary due to the nature of evolutionary dynamics where fluctuations matter even in large populations. Their heuristic solution they present is of the form of a bang-bang control. In this presentation, I reformulate the problem as a path integral control problem where a more exact treatment is possible. I will give an introduction to stochastic differential equations and the path integral control theory. Subsequently, I will outline how to compute the adaptive optimal control by sampling.
      Speaker: Bert Kappen
    • 39
      Optimal evolutionary control of molecular phenotypes
      Speaker: Armita Nourmohammad
    • 40
      Speaker: Ceyhun Eskin
    • 41
      Speaker: Colin LaMont
    • 42
      A bottom-up approach to microbial evolution and ecology
      Speaker: Jeff Gore
    • 43
      Plastic tradeoffs in evolution: a simple theoretical model
      Performance tradeoffs are fundamental to both ecology and evolution, but in most models, are simply postulated. This approach is justified for tradeoffs enforced by rigid biophysical or biochemical constraints; unsurprisingly, some of the best-understood examples are in this class. However, experimental results suggest that many relevant tradeoffs are not rigid, but can themselves evolve. I will describe a simple theoretical framework for studying how an evolving tradeoff structure both shapes and is shaped by the evolutionary trajectory, offering a promising path towards controlling or predicting evolution. I will discuss some unexpected behaviors predicted to arise as a result of this feedback loop.
      Speaker: Mikhail Tikhonov
    • 44
      Speaker: Claudia Bank
    • 45
      Speaker: Oskar Hallatscheck
    • 46
      Emergence and propagation of epistasis in metabolic networks
      The effect of a mutation on a phenotype of interest often depends on the presence of other mutations in the genome. Such dependencies are known as epistasis or genetic interactions. The evolutionary process fundamentally depends on the structure and type of these interactions. Certain types of epistasis are involved in explaining the evolution of sexual reproduction, historical contingency, robustness to deleterious mutations, etc. Epistasis is also extensively used in genetics to identify genes involved in various biological processes. Despite its prominent role in biology, our current understanding of the mechanistic origins of epistasis is poor, especially for mutations affecting different genes. In particular, we lack a null expectation for what types of epistasis (if any) should be common to many or even all biological systems and what types of epistasis may be signatures of potentially interesting idiosyncratic interactions between specific gene products. Here, I develop a mathematical theory for understanding what types of epistasis we might expect to observe between mutations affecting microbial metabolism. I consider a hierarchy of increasingly coarse- grained descriptions of a metabolic network, such that more coarse- grained (“higher-level”) descriptions typically have fewer effective parameters than more detailed (“lower-level”) descriptions, with the growth rate being the single top-level parameter. I find that mutations that exhibit no epistasis for lower-level parameters (e.g., mutations affecting different enzymes) almost certainly exhibit epistasis for higher-level parameters, and that epistasis for lower-level parameters generically implies epistasis for higher-level parameters. This suggests that any metabolic mutations that have effects on growth rate are generically expected to exhibit epistasis. Moreover, I show that, for networks with first-order reaction kinetics, negative epistasis at a lower level remains negative at all higher levels, and strong positive epistasis at a lower level remains strongly positive at all higher levels. Finally, I show that certain topological relationships between reactions within the network impose constraints on the sign of epistasis for growth rate that mutations affecting these reactions can exhibit. This theory provides a foundation for interpreting epistasis observed in experiments and for constructing more realistic models of genome-wide fitness landscapes.
      Speaker: Sergey Kryazhimzkiy
    • 47
      Interrogating evolution using robotics-assisted continuous evolution
      Directed evolution provides a platform to study evolution in the laboratory. Previously, Phage Assisted Continuous Evolution (PACE) has been used to measure the stochasticity of evolution results and interrogate the impact of evolutionary parameters such as mutation rate and selection stringency. Although this technique allows for many generations to be observed in laboratory timescales (1 generation per 20 minutes), the experiments difficult to multiplex. To address this issue, we have implemented a robotic platform that enables massively multiplexed continuous directed evolution. Currently, our platform can implement 96 parallel evolution experiments, monitor a luminescence or fluorescence readout of fitness in real time, and vary the evolutionary conditions (such as mutation rate) amongst these experiments. We plan to use this platform to investigate the impact of cycling between different evolutionary conditions on the outcomes of directed evolution experiments.
      Speaker: Erika DeBenedictis
    • 48
      Evolution in fluctuating environment
      All environments vary in time, and these fluctuations affect the survival and reproduction of individuals. Uncorrelated fluctuations among individuals produce demographic stochasticity (genetic drift). Coherent fluctuations across entire populations (environmental stochasticity) produce random fluctuations in allele fitness and in carrying capacity. In some cases, and in particular under the standard Moran dynamics, fluctuating fitness acts as a balancing selection mechanism and stabilizes an attractive coexistence state. Noise induced stabilization (NIS) of this type affects dramatically the evolutionary dynamics. I will present a few results that have to do with the evolutionary process in fluctuating environment, with and without NIS. These include: 1. The chance of ultimate fixation, the time to fixation and the time to absorption (either fixation or loss) for a two-allele system under drift, fluctuating selection and fluctuating carrying capacity [1-3]. 2. NIS assisted stochastic tunneling [2]. 3. Site-frequency spectrum for multi-allele system [4]. 4. The speed of evolution under fluctuating selection in the successional-fixation phase and in the clonal interference phase [5]. 5. Inference problems, the effect of environmental stochasticity on dn/ds ratio, McDonald-Kreitman test, coalescence theory and so on. [1] Fixation and absorption in a fluctuating environment. M Danino, NM Shnerb, Journal of theoretical biology 441, 84-92 (2018). [2] Noise-induced stabilization and fixation in fluctuating environment, I Meyer, NM Shnerb, Scientific reports 8 (1), 9726 (2018). [3] Stability of two-species communities: drift, environmental stochasticity, storage effect and selection, M Danino, DA Kessler, NM Shnerb, Theoretical population biology 119, 57-71 (2018). [4] Theory of time-averaged neutral dynamics with environmental stochasticity, M Danino, NM Shnerb, Physical Review E 97 (4), 042406 (2018) [5] Environmental stochasticity and the speed of evolution, M Danino, DA Kessler, NM Shnerb, Journal of Statistical Physics 172 (1), 126-142 (2018)
      Speaker: Nadav Shnerb
    • 49
      Poster Flash Talks W3
    • 50
      Poster session W3
    • 51
      Antibody-mediated cross-linking of gut bacteria hinders the spread of antibiotic resistance
      The body is home to a diverse microbiota, mainly in the gut. Using stochastic models of bacterial population dynamics, we contributed to show that the main physical effect of a type of antibodies which are the main effector of the adaptive immune response secreted in the gut, is to cross-link bacteria into clusters as they divide, preventing them from interacting with epithelial cells, thus protecting the host. This yields clonal clusters of bacteria, which could impact the diversity of the bacterial population, and thus adaptation. Resistant bacteria are selected for by antibiotic treatments, and once resistance becomes widespread in a population of hosts, antibiotics become useless. Here, we develop a multiscale model of the interaction between antibiotic use and resistance spread in a host population, focusing on this important aspect of within-host immunity. We demonstrate that immunity-driven bacteria clustering can hinder the spread of a novel resistant bacterial strain in a host population. We quantify this effect both in the case where resistance pre-exists and in the case where acquiring a new resistance mutation is necessary for the bacteria to spread. We further show that the reduction of spread by clustering can be countered when immune hosts are silent carriers, and are less likely to get treated, and/or have more contacts. We demonstrate the robustness of our findings to including stochastic within-host bacterial growth, a fitness cost of resistance, and its compensation. Our results highlight the importance of interactions between immunity and the spread of antibiotic resistance, and argue in the favor of vaccine-based strategies to combat antibiotic resistance.
      Speaker: Claude Loverdo
    • 52
      High mutation rate evolution during adaptation to high salinity or sub-inhibitory concentrations of gentamicin
      The drastic increase in mutation rate due to hyper-mutator genotype has been shown to be favoured when a population is adapting to stressful conditions. This is because they allow a faster provision of beneficial mutations and hitchhike with the adaptive mutations they have triggered. Here, we use an Escherichia coli strain with a high probability of becoming hypermutator to investigate whether the nature of the stress and its intensity have an effect on the selection of hypermutator genotypes. To do so, we experimentally evolved E. coli populations across a gradient of salinity and of subinhibitory concentration of gentamicin. We show that the nature of the stress and the associated target size for adaptive mutation strongly influences the probability for a population to become hyper-mutator. We then decipher the similarities and differences in the molecular bases of adaption for hyper-mutator and wild-type genotypes.
      Speaker: Stephanie Bedhomme
    • 53
      Dimension-Reduction Theory for Direction and Constraint in Phenotypic Evolution
      Directionality and constraint in phenotypic evolution in terms of phenotypic fluctuation and response against environmental change are discussed. From dynamical-systems theory and evolution simulations, we first demonstrate link between robustness to noise and to mutation. This leads to proportionality between phenotypic plasticity by genetic change and by environmental noise, which implies correlation between plasticity by environmental perturbations and genetic change. Next, global proportionality in the changes of high-dimensional phenotypes (say expression of thousands of levels) due to environmental and evolutionary changes is uncovered from simulations and experiments. All the observed statistical laws for cellular states are explained by a theory that adaptive changes in high-dimensional phenotypes are restricted within a low-dimensional slow manifold, as a result of dynamic robustness and evolutionary plasticity. The present results are also confirmed by simulations of gene-expression and catalytic reaction dynamics, as well as statistical physics model of evolving interacting spins. Connection with bacterial evolution experiments as well as protein data is also discussed. As related topics, I also hope to discuss evolution-development congruence, origin of central dogma as a result of symmetry breaking in function and information 1. Kaneko K., Life: An Introduction to Complex Systems Biology, Springer (2006) 2. K. Sato, Y,Ito, T.Yomo, K. Kaneko, "On the Relation between Fluctuation and Response in Biological Systems" Proc. Nat. Acad. Sci. USA 100 (2003) 14086-14090\\ 3. K. Kaneko, "Evolution of Robustness to Noise and Mutation in Gene Expression Dynamics" PLoS One(2007) 2 e434 4. K. Kaneko, "Phenotypic Plasticity and Robustness: Evolutionary Stability Theory, Gene Expression Dynamics Model, and Laboratory Experiments", Evolutionary Systems Biology (2012) (Springer, ed. O. Soyer) 5. K. Kaneko, C.Furusawa, T. Yomo, "Macroscopic phenomenology for cells in steady-growth state", Phys.Rev.X(2015) 011014 6. C. Furusawa, K. Kaneko "Global Relationships in Fluctuation and Response in Adaptive Evolution", J of Royal Society Interface (2015) 7. K. Kaneko, C. Furusawa, “Macroscopic Theory for Evolving Biological Systems Akin to Thermodynamics”, Annual Rev. Biophys. (2018) 47, 273-290 8.T Kohsokabe, K Kaneko, Evolution‐development congruence in pattern formation dynamics: Bifurcations in gene expression and regulation of networks structures - Journal of Experimental Zoology B 326 (2016)61 9.The origin of a primordial genome through spontaneous symmetry breaking N Takeuchi, P Hogeweg, K Kaneko - Nature Comm 8(2017)250 10 N Takeuchi, K Kaneko,”The origin of the central dogma through conflicting multi-level selection”, bioRxiv, 2019
      Speaker: Kunihiko Kaneko
    • 54
      Welcome W4
    • 55
      Physics and evolution of long-range effects in proteins
      Proteins are very heterogeneous objects: they are sensitive to perturbations at some sites distant from their active site while being insensitive to perturbations at closer sites. These long-range effects make decoding and engineering protein sequences particularly challenging. I’ll review the evidence for the ubiquity of these long range effects and discuss previous models aimed at understanding their physical nature and evolutionary origin. This will motivate the introduction of a new model where long-range effects emerge spontaneously. I’ll explain how the model accounts for the evolution of long-range regulation (allostery) and for the different patterns of coevolution that may be inferred from protein sequences.
      Speaker: Olivier Rivoire
    • 56
      Phenotype evolution as optimization
      Speaker: Jakub Otwinowski
    • 57
      The sequence-affinity landscape of antibodies
      Speaker: Alexandra Walczak
    • 58
      Comprehensively mapping phenotypic tradeoffs and epistasis using microscopy-based deep mutational scanning at the million-mutant scale
      The difficulty of predicting the course of evolution, even on the single- protein scale, stems from: (1) the immense size of sequence space on which evolutionary trajectories lie; (2) the fact that fitness may be a complex function of many phenotypes; and (3) phenotypic heterogeneity, which implies that the fitness of a particular genotype is not solely determined by its mean phenotypes, but by its distribution of phenotypes. To collect the data requisite to have a hope of predicting evolution, therefore, one needs a technique which offers extremely high throughput measurements of multiple phenotypes at the single-cell level. We have recently developed the capability to phenotype a million bacterial strains per day using “mother machine” microfluidic chips, fast fluorescence microscopes, and a petabyte-scale data analysis pipeline. We are working towards producing a genotype-to-phenotype map of all 11 million double mutants of GFP, measuring brightness, maturation kinetics, photobleaching kinetics, propensity to aggregate, and two-dimensional emission-excitation spectra. We are also measuring induction curves for millions of double mutants of lacI repressor. These datasets will, for the first time, comprehensively reveal how different biophysical properties trade off against each other in a local fitness neighborhood of a protein. These kinds of data will allow engineering variants of these proteins with specific desired properties; more generally, they will inform theoretical models of protein evolution dynamics by characterizing epistasis and tradeoffs in real proteins.
      Speaker: Jacob Shenker
    • 59
      Inferring large cooperative patterns
      While they were statistically shown to be present in evolutionary data, large cooperative units, sometimes referred to as "sectors" have not be observed in Direct Coupling Analysis (DCA) of protein multiple sequence alignment. We show that due to the sampling limitations of evolutionary trajectories, these units are bound to be highly conserved and as such invisible to analysis that is focusing on correlations. We propose a practical way to quickly augment the DCA procedure that would allow for discerning the sectors by factoring in the conservation pattern, similar to the Statistical Coupling Analysis (SCA), which was the original method of observing cooperative units.
      Speaker: Yaakov Kleeorin
    • 60
      Affinity maturation of antibodies and the puzzle of HIV low spike density
      Affinity maturation (AM) is the process through which the immune system evolves antibodies (Abs) which efficiently bind to antigens (Ags), e.g. to spikes on the surface of a virus. This process involves competition between B-cells: those that ingest more Ags receive signals (from T helper cells) to replicate and mutate for another round of competition. Modeling this process, we find that the affinity of the resulting Abs is a non-monotonic function of the target (e.g. viral spike) density, with the strongest binding at an intermediate density (set by the two-arm structure of the antibody). We argue that, to evade the immune system, most viruses evolve high spike densities (SDs). This is indeed the case, except for HIV whose SD is two orders of magnitude lower than other viruses. However, HIV also interferes with AM by depleting T helper cells, a key component of Ab evolution. We find that T helper cell depletion results in high affinity antibodies when SD is high, but not if SD is low. This special feature of HIV infection may have led to the evolution of a low SD to avoid potent immune responses early on in infection.
      Speaker: Mehran Kardar
    • 61
      Coevolution in the adaptive immune system
      Speaker: Armita Nourmohammad
    • 62
      Antibody affinity maturation as a model system for sequence evolution
      Stochastic drift and adaptive determinism together shape populations of functional sequences. Untangling these forces by studying natural populations is challenging because a given evolutionary story is told only once. We use an evolutionary system that can be run in replicate —somatic evolution of antibody repertoires in jawed vertebrate immune systems—as the context for tractable models that quantitatively characterize the roles of contingency and determinism. Using data from experimental model systems of increasing biological realism, we develop theoretically-motivated computational methods to characterize the evolutionary dynamics of antibody affinity maturation.
      Speaker: William DeWitt
    • 63
      Does evolution care about bits?
      Given the stochastic nature of gene expression, genetically identical cells exposed to the same environmental inputs will produce different outputs. This heterogeneity has consequences for how cells are able to survive in changing environments. Recent work has explored the use of information theory as a framework to understand the accuracy with which cells can ascertain the state of their surroundings. Yet the predictive power of these approaches is limited and has not been rigorously tested using precision measurements. To that end, we generate a minimal model for a simple genetic circuit in which all parameter values for the model come from independently published data sets. We then predict the information processing capacity of the genetic circuit for a suite of biophysical parameters such as protein copy number and protein-DNA affinity. We compare these parameter- free predictions with an experimental determination of the information processing capacity of E. coli cells, and find that our minimal model accurately captures the experimental data. These theoretical results will allow us to tackle the question of the abstract quantity of information bits can serve as a quantitative trait on which selection can act.
      Speaker: Manuel Razo Mejia
    • 64
      The energetics of molecular adaptation: Mutations, epistasis, and allostery in transcriptional regulation
      Quantitative explorations of evolutionary process often rely on the generation of fitness landscapes frequently depicted as two- dimensional surfaces upon which peaks and valleys in fitness are given as functions of phenotypes. However, we are often unable to quantitatively describe these phenotypes in experimentally accessible terms, hindering our ability to test these predictive models. This is especially true in the context of gene expression where the rich phenomenology of gene expression dose-response curves (such as leakiness, dynamic range, and sensitivity of response) are left to qualitative or semi-quantitative description through the use of Hill functions. Here, we present a general theory of allosteric transcriptional regulation using the Monod-Wyman-Changeux model and derive an expression for the free energy of an allosteric repressor. We rigorously test this model using the simple repression motif in bacteria by predicting and measuring the behavior of strains that span a swath of repressor copy numbers and DNA binding strengths. Our model captures the induction profiles of these strains and generates analytic expressions for key properties such as dynamic range and [EC50]. Furthermore, we explore how mutations at the level of amino acid substitutions are connected to the various biophysical parameters which govern the response of the system and generate predictions of how each mutation modulates the free energy of the repressor. With an array of well characterized point mutants of the repressor, we test the predictions of how the pairwise double mutants behave, revealing that in most cases the energetic contributions of each individual mutation are additive. Finally, we explore how epistatic interactions could be manifest in our model.
      Speaker: Griffin Chure
    • 65
      Poster Flash Talks W4
    • 66
      Poster Session W4
    • 67
      Identifying responding clonotypes in immune repertoires
      Speaker: Aleksandra Walczak
    • 68
      Methods for identification of condition-associated T-cell receptor clonotypes in RepSeq data
      High throughput sequencing of antigen receptor repertoires (RepSeq) allows for the sequencing of millions of TCR/BCR sequences per sample. However, our ability to extract clinically relevant information from repertoire sequencing data is still limited. Here we present three computational approaches to identify vaccination, infection, cancer, or autoimmune disease-associated clonotypes from longitudinal RepSeq data (several time points after treatment for one donor), population RepSeq data (repertoires from patient cohort) and single repertoire samples. First, we present a statistical model which detects clonal expansion by analyzing TCR cDNA count data from quantitative repertoire sequencing. We applied this model to TCRbeta repertoires of three twin pairs after yellow fever immunization (YFV17D). We identified 500-1500 expanded clonotypes in each donor and validated them for YFV17D-specificity by three independent functional assays. Second, we describe an algorithm to identify disease-specific clonotypes using repertoires from cohorts of patients. A stochastic model of TCR recombination is used to identify clonotypes shared between more substantial numbers of patients than one could expect by chance. This extensive sharing could only be explained by clonal selection to same antigens on the periphery. Using this model, we identified known public cytomegalovirus and ankylosing spondylitis- associated clonotypes in the respective patient cohort repertoires.vThird, we extended this approach from the population level to single samples. We clustered TCRbeta clonotypes by significant sequence convergence, which is estimated and weighted from a stochastic TCR generation model. We show that identified clusters are abundant after YF-immunization and consist predominantly of YF- specific clones and almost absent before immunization. These three approaches allow the identification of disease-specific TCR variants using sequencing data only.
      Speaker: Mikhail Pogorelyy
    • 69
      Alterations in TCR repertoire after yellow fever vaccination and revaccination
      Yellow fever vaccination is a well-established model of acute viral infection in human. Primary immunization elicits strong T-cell response and formation of long-lived memory. However, little data exists on T- cell response to revaccination. We applied deep TCR-profiling to track T-cell clones after yellow fever vaccination and revaccination. We isolated PBMCs, CD4+, CD8+, memory and MHC-dextramer positive subpopulations from the first-time vaccinee and revaccinated donor on several time points before and after YFV17D immunization. TCR repertoires were reconstructed, and expanded TCRs were identified using edgeR. We identified 1580 expanded clonotypes in first-time vaccinee and 204 in the revaccinated donor. YFV17D-specificity was confirmed by sequencing of MHC-dextramer positive subpopulation. In first-time vaccinee fraction of responded clones sharply peaked (6.1%) at day 15 after immunization. However, in revaccinated donor fraction of responded clones reached 0.4% by day 5 and slightly increased on days 10 and 15. Analysis of individual clones dynamics showed the presence of two distinct groups of clones in the revaccinated donor. T- cell response to booster YFV17D vaccination, even 30 years after the first, differs from response to primary immunization both by intensity and dynamics. We detected the early expansion of some T-cells in response to revaccination and delayed response of putatively naive T- cells.
      Speaker: Asya Minervina
    • 70
      Prediction in immune repertoires
      Speaker: Thierry Mora
    • 71
      Optimal Memory Strategies in the adaptive immune system
      The adaptive immune system in vertebrates mounts a specific response against a multitude of distinct pathogens by generating highly diverse immune (B- and T-cell) receptors. After battling with an infection, the immune system keeps a memory of its effective immune receptors as a token to act quicker in future encounters. Our goal is to understand the statistical principles of such memory formation through modeling of the immune response as a decision-making process [1]. We relate the adaptation time of an antibody resembling B-cell receptor to a temperature-like quantity that sets the limit on the decision-making prior. Based on the concept of out-of-equilibrium decision-making [2] we derive equilibrium and out-of-equilibrium memory strategies and analyze their utility. [1] Pedro A. Ortega and Daniel A. Braun. Thermodynamics as a theory of decision-making with information-processing costs. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences [2] Jordi Grau-Moya et al. Non-equilibrium relations for bounded rational decision-making in changing environments. Entropy
      Speaker: Oskar Schnaack
    • 72
      Featureless selection of informative neurons in the brain using MultiScale Relevance
      How a neuron responds to complex stimuli, behaviors, and tasks can encompass a wide range of time scales. Understanding how information is represented in these responses across multiple temporal resolutions then requires measures that go beyond imposing symmetry constraints on the neuron’s tuning curves. In this study, we propose a non- parametric, model-free indicator – which we call multiscale relevance (MSR) – to quantify the dynamical variability of neural spiking across multiple time scales. This allows us to select relevant neurons using only the time stamps of the spiking activity without resorting to any a priori external covariate or any specific symmetries in the neurons’ tuning curves. This fully featureless selection is done by identifying neurons that have broad and non-trivial distribution of spike frequencies across a broad range of time scales. When applied to neural data from the medial entorhinal cortex, and from the thalamic and post-subicular regions of freely- behaving rodents, we found that neurons having low MSR tend to have low mutual information and low firing sparsity across the external correlates that are believed to be encoded by the region of the brain where the recordings were made. In addition, neurons with high MSR contain significant information on spatial navigation and allow to decode spatial position or head direction as efficiently as those neurons whose responses have high mutual information with the covariate being decoded. With these results, we propose that the MSR can be used as an unsupervised method to rank and select information-rich neurons from a heterogeneous population without the need to appeal to any a priori external covariate.
      Speaker: Ryan Cubero