Magnus Nordborg: Genetic architecture in the presence of selection

This sub-project is primarily concerned with how we can determine the genetic architecture of traits that have been under selection, which is difficult because selection, almost by definition, causes statistical associations (“linkage disequilibrium”) between causal loci, making it difficult to estimate the effect of individual loci. We will investigate how standard GWAS methods are affected by selection, and develop methods that take selection into account. Both real and simulated data will be analyzed - in close collaboration with the other participants of the SFB.

Kelly Swarts: Adaptation to dynamically changing environments in Norway spruce (Picea abies) forests

We focus on Norway spruce (Picea abies) as a model for understanding the genetics underlying adaptation to local environment and competition in a long-lived organism. Annual growth rings (tree-rings) in temperate tree species are a unique resource for directly measuring annual growth over the life a tree, allowing us to estimate genotypic performance across years with independently assessed environmental data. Growth is a complex, integrative trait, and parsing this system using regression modeling approaches allows us to evaluate plastic responses to changing environment (GxE) and, in conjunction with genomic estimates, adaptation to specifically modeled environments. Environments experienced are multi scalar and composed of both biotic and abiotic components. We focus here on assessing adaptation to competition, or the biotic intra-specific environment, which preliminary results show to change over the course of a tree’s life.

Nick Barton: Understanding polygenic adaptation, despite polygenic selection and ancestral noise

Genetic lineages trace far back into the past, and so present-day sequences are shaped by obscure processes. This sub-project asks how we can separate the effects of selection from the ancestral “noise” that arises from a complex population structure. The aim is to develop methods for inference from both experimental and natural populations, and in parallel, to better understand how selection can successfully disentangle adaptive variants despite linkage. Current methods are primitive, in that they largely look at sites separately. In contrast, here we follow the descent of blocks of genome through the population. Such ideas have only begun to be developed very recently, stimulated by the availability of long-read sequence data, and computational innovations. We combine mathematical analysis, using branching processes and generating functions to describe linked blocks, with simulations of continuous genomes. Methods will be tested on experimentally selected populations (where ancestors are known), and also on data from nature (e.g. Antirrhinum, Arabidopsis, and Picea).

Joachim Hermisson: Polygenic adaptation beyond sweeps and shifts

Molecular population genetics and quantitative genetics have developed opposite views of the adaptive process, invoking either selective sweeps or small polygenic allele frequency shifts. While these narratives clearly refer to different regions of biological parameter space, it is less clear how these regions are demarcated or how polygenic adaptation looks like in between. The aim of this sub-project is to shed light on this transition range. We will use analytical modeling, complemented by large-scale computer simulations to characterize adaptive architectures beyond sweeps and shifts. Based on these results, we will develop statistical methods to estimate key parameters that determine the adaptive scenario from replicated time-series data of adaptive alleles and from polygenic footprints in DNA diversity data. We will place a special focus on model scenarios that are tailored to the empirical E&R and GWA studies of the SFB.

Himani Sachdeva: Can the infinitesimal model reconcile observations from GWAS and E&R?

The goal of this project is to understand whether one can reconcile heterogeneous selection response along the genome (as frequently observed in E&R) with highly polygenic architectures of selected traits (as suggested by GWAS) under the novel ‘infinitesimal model with linkage’. We will investigate how selection on a large number of infinitesimal genetic variants on a linear genome distorts genealogies along the genome, and whether this results in signatures of selection that can be detected in timeseries of allele frequencies or snapshots of sequence variation (e.g. sweep-like patterns). Another goal is to understand how different kinds of statistical associations (LD) between infinitesimal variants - LD present in the founding population at the onset of selection versus LD that builds up during selection response - influence heterogeneity of response across the genome and across replicates. The projects will employ a range of computational and mathematical approaches.

Neda Barghi: Genetic and adaptive architecture of polygenic traits

This project will characterize adaptation of a polygenic trait under stabilizing selection after a shift in its optimum. We will perform an experiment where replicates of a Drosophila simulans population adapt to a new optimum (larger female body size) using a pre-defined fitness function. By generating a comprehensive genomic and phenotypic dataset, we will identify the alleles that respond to selection, i.e. adaptive architecture, and provide empirical data for adaptive processes of polygenic traits. We will compare this adaptive architecture to the one inferred under truncating selection. We will work closely with theoreticians in the SFB to use the inference methods developed in these projects to test for selection. The alleles that underlie body size, however, will outnumber the ones identified in adaptive architecture. So, to identify alleles with adaptive potential and constraints we will characterize genetic architecture of female body size by GWAS and compare it with adaptive architecture.

Robert Kofler: Genetic and adaptive architecture of body size variation under truncating selection in Drosophila

Due to constraints acting on loci, such as linkage to unconditionally deleterious alleles or pleiotropic effects, the loci contributing to phenotypic variation are not necessarily identical to the loci responding to selection. We will test this hypothesis by unraveling i) the genetic basis of female body size in D. simulans - a highly polygenic trait - with a GWAS, and ii) the adaptive architecture under truncating selection with an Evolve and Resequence study (E&R), where the largest females are selected at each generation. Discrepancies between the genetic and the adaptive architecture will allow us to identify the extent of constraints acting on the genetic basis of a trait and to identify variation of these constraints among the loci contributing to the trait. Furthermore, we will test whether an E&R study with an optimized selection is more powerful for finding the genetic basis of a trait then a GWAS, as predicted by our previous simulation study. By contrasting our results to stabilizing selection experiments we will test whether the adaptive architecture differs between truncating and stabilizing selection.

Christian Schlötterer: Polygenic adaptation with reduced genetic variation

Using a large number of replicate populations derived from a cross between two isogenic Drosophila lines, this project exploits the stochastic nature of recombination to generate haplotype blocks with different configurations of loci contributing to a new extreme phenotypic optimum (29°C). During the early phase of the experiment, only few recombination events occur and large genomic regions will respond to selection. At later generations, increasing numbers of recombination events combine favored alleles from both strains. Following the frequency trajectories of strain-specific alleles, such recombination events can be identified by frequency change in opposite direction. The stochastic nature of recombination will result in heterogeneity among replicates, which in turn provides information about the clustering of loci contributing to the phenotype. It is anticipated that this project provides information about key parameters of the adaptive architecture – the number of selection targets, their effect sizes and distribution over the chromosome.