Julien F. Ayroles (organiser), Benjamin de Bivort, Örjan Carlborg, Andrew G. (Andy) Clark, Fabien Duveau, Ian Dworkin, Marie-Anne Félix (organiser), Greg Gibson, Arthur Ko, Magnus Nordborg, Virginie Orgogozo, Annalise Paaby, Päivi Pajukanta, Stephen (Steve) Proulx, Christine Queitsch, Scott Rifkin, Mark L. Siegal, Henrique Teotonio, Gaël Yvert, Noah Zaitlen
by Julien Ayroles and Marie-Anne Felix
4 – 9 July, 2016
The current quantitative genetic paradigm, devised by RA Fisher in the 1930s, is driven by a prevailing view that additive genetic models, focused on the mean effect of alternative alleles, explain variation in most phenotypes. The recent development of powerful genomic technologies enabled the search for these additive effects across entire genomes and quickly led to the popularization of genome-wide association studies (GWAS). Unfortunately, a decade later and in spite of much effort, we have fallen short of the goal of explaining most of the heritability for complex traits in terms of allelic effects. Given sufficiently large sample sizes, there is little doubt that GWAS can identify a large proportion of marginal allelic contribution to complex traits (and this provides a goldmine of information for biologists interested in the functional role of genes). However, this approach is designed to describe the average effect of an allele randomized over a large number of the genetic backgrounds and environments. A fast-growing body of evidence indicates that the genotype-phenotype map is much more complicated than Fisher’s additive model would predict. While this is the currently the focus of much debate, complex, non-additive interrelationships between loci appear to be the rule and not the exception, and these allelic effects are often environmentally sensitive. In that context, the paradigm derived from traditional quantitative genetics is at odds with our goals because we often seek to understand the causal path from genotype to phenotype for individuals and not populations. When the average allelic effect does not capture a specific allelic effect under certain conditions, whether due to genetic background or environment, current methods will not detect the link between genotype and phenotype. Historically, the genetic architecture of complex traits has been treated as a blueprint of the phenotype, with a static set of parameters. This framework has been useful in breeding programs and estimating heritability, however this conception may be misleading because the blueprint itself is context dependent and highly dynamic.
Keywords: Variability, plasticity, epistasis, quantitative genetics, robustness, decanalization
This meeting centered on addressing a set of key questions:
– How can we reconcile the high dimensionality of genomics with our drive to understand the determinant of phenotypic variability at the molecular level and with a mechanistic appreciation of how variation in genetic background and/or environmental variation modify allelic effects?
– It is well established in quantitative genetics that stressful environmental exposure tends to increase the phenotypic variance of a population, but how and why? And why do some individuals appear to be more sensitive than others to such perturbations?
– Shifting Focus From Mean To Variance: how to investigate the roles robustness plays in driving either disease emergence or adaptive change.
These questions are central to anyone interested in understanding the genetic basis for complex trait variation both from a genetic architecture perspective and at the functional level. Addressing them requires a broad set of perspectives, and this was the spirit of this meeting. Participants came from a wide range of backgrounds: developmental biology, computer science, quantitative and population genetics, statistics, neuroscience and mathematical biology. Here are a few highlights.
Marie-Anne Felix presented a fascinating model to study developmental homeostasis: the fate of the Caenorhabditis vulva precursor cell. Using this model, she is able to study the effect of random genetic variation on phenotypic variation. Specifically, she is able to determine that random genetic variation affects the fate of one specific cell more than the others and that for the latter, cryptic genetic variation is a critical component of this system that maintains the correct number and position of vulva cells. Regulating variability is a fundamental factor for development and this work elegantly connect molecular mechanisms with evolutionary mechanisms.
Gael Yvert presented a set of thought provoking ideas that have the potential to redefine how we approach QTLs. Traditionally QTLs have been associated with average difference between genotypes. However, many studies have now shown that variability itself, as a trait, is under genetic control and that QTL affect variance. This adds an important layer of regulation underlying inter-individual phenotypic differences. Working with yeast (a system where many single cells of a given genotype can be phenotype) he is able to show that different genotypes exhibit substantial phenotypic individuality and map the genetic determinants of variability. Under this model, genotypes do not control the mean but a distribution or the probability that a given cell/individual to land at a given point in phenotypic space. This work has major implications for anyone thinking about personalized medicine and prediction from genotype to phenotype.
Andrew Clark and Steve Proulx took us back to the fundamentals. Under what evolutionary scenarios can variance controlling alleles evolve? They presented both theoretical and experimental approaches to address this question.
Greg Gibson, Noah Zaitlen and Päivi Pajunkanta brought the problem of variability and predictability to human health. Globally, we are witnessing the rise of a variety of complex diseases related to a dramatic change in our environment (e.g. our diet in the western world). Conditions such as obesity, asthma, depression, and type-2 diabetes are part of a long list of non-infectious diseases that are largely determined by a complex interplay between the genetic make up of an individual and his or her environment. Their approach has built upon a strong foundation of evolutionary theory indicating that organisms must maintain some degree of phenotypic conformity in light of considerable environmental and genetic perturbation, a process C. H. Waddington called canalization. Because biological systems evolve to maintain homeostasis under a certain range of environmental (or genetic) perturbations, changing the environment outside the range of what organisms commonly experience will lead to a shift in an optimum that has likely been shaped by stabilizing selection over many generations. This provides a framework from which to study how the rapid transition from our ancestral environment to our modern lifestyle could have lead to the current epidemic of common genetic diseases.
Scott Rifkin’s presentation highlighted a common theme during our meeting – it is all about the phenotype! New technologies are constantly pushing the boundaries of what we can see and measure and the more we can see, the more we understand the importance of robustness and variance control. Scott presented work in yeast, where he is able to visualize mRNA molecules in real time, in vivo, in single cells. This allows him to study the dynamics and variance in transcriptional regulation at an unprecedented scale with unprecedented resolution.
Lastly, Ben de Bivort presented an experimental model to study the neuronal basis for behavioral individuality in Drosophila. Using individual chambers, he is able to measure a wide range of behaviors (e.g. activity, odor preference, phototaxis, etc). Holding genotype and environment constant, he is able to show that there is substantial amount of intra-genotypic variability among genetic backgrounds. In at least one instance (locomotory handedness) he was able to map a brain region controlling behavioral variability – once again pointing to the importance of studying variability as a trait.
The difficulty faced by the genetic community to describe the genetic basis of complex traits underscores a fundamental knowledge gap in our understanding of the context-dependency of allelic effects. This meeting centered on the notion that variability is a fundamental property of biological systems: it is regulated, it evolves and when disrupted often leads to aberrant phenotypes (e.g. diseases). To study the connection between genotype and phenotype, we should go beyond the standard approaches focused on the average phenotypic effect between alternative alleles. Instead, we should start thinking about how alternative genotypes are associated with different degrees of phenotypic variability, a process regulated throughout ontogeny.
This meeting was without a doubt one of the most rewarding and intellectually stimulating many of us have attended. Doing away with computer-based presentations and going back to the black board was particularly conducive to very lively exchanges. Participants covered the full range from advanced Ph.D student (1), to postdoc (2), junior faculty (8) and senior faculty (10).