Annual Retreat 2023

September 8th, Friday @ Monadnock Room (2nd fl), Merkin Building, Broad Institute (415 Main Street, Cambridge, MA 02142)

- Registration is required -

Registration QR


8:50AM - Breakfast


9:10AM - Welcome/Introduction [Organizers]


9:20AM - *Keynote Speaker : Magnus Nordborg

:: Scientific Director, Gregor Mendel Institute of the Austrian Academy of Sciences ::

  • Title: Field GWAS and experimental evolution experiments reveal genetic tradeoffs in response to temporally and spatially variable selection
  • Summary: Local adaptation has been demonstrated in many organisms, but the traits involved, and the temporal and spatial scales at which selection acts are generally unknown. We carried out a multi-year study of 200 natural inbred lines of Swedish \emph{Arabidopsis thaliana}, using local field sites, and common-garden experiments that primarily measured fertility, as well as evolution experiments that measured fitness over the full life cycle. The latter revealed a massive fitness advantage to large-seeded beach lines, which presumably had an advantage during seedling establishment, but which also had low fertility. Southern lines had high fertility in most sites and years, but were sensitive to harsh winters in northern sites, and also showed less resistance to sporadic heavy slug herbivory. Rather than a simple pattern of local adaptation, we found evidence of strong temporally and spatially varying selection on multiple traits, with trade-offs at every level.

10:30AM - Coffee Break (30min)


11:00AM - 1) Olivia Harringmeyer: Hopi Hoekstra lab

  • Title: The evolution of an inversion supergene in deer mice
  • Summary: How chromosomal inversions arise and become supergenes - linking multiple adaptive mutations - remains a major open question in evolutionary biology. Here, we identified a massive (40-megabase) inversion supergene in deer mice that is strongly associated with longer tails and darker coat colors, two locally adaptive traits in forest mice. To understand the evolution of this inversion supergene, we isolated the inversion on a consistent genomic background and investigated its specific effects on both traits. Using RNA-sequencing, genetic mapping and comparative genomic approaches, we identified top candidate mutations within the inversion for both tail length and coat color. Further, we found that this inversion became a supergene through the accumulation of adaptive mutations over time, rather than at its initial formation. Together, our study suggests that inversions can be an important mechanism facilitating local adaptation through linking together adaptive mutations that arise within the inversion over time.

11:20AM - 2) Trevor Cousins: David Reich lab 

  • Title:  Investigating the genetic relationship between modern humans, Neanderthals and Denisovans
  • Summary:

    "The standard model of hominin population history has become complicated. It is one in which Neanderthals, Denisovans, and modern humans are related by at least four admixture events prior to the introgression of Neanderthals and Denisovans into modern humans <60kya. However, the standard model (see below) has to go through contortions to explain Y chromosome and mitochondrial DNA. Specifically, the autosomal relationship suggests Neanderthals and Denisovans are a clade and a sister group to modern humans, but the mitochondrial and the Y chromosome data suggest that Denisovans are the outgroup. Moreover, the inferred split times from each data type between these groups are not well aligned: between modern humans and Neanderthals, the mitochondrial DNA, Y chromosome, and autosomes suggest split times of 470-360kya, 410-320kya, and 550-770kya, respectively. This would imply that both the Neanderthal Y chromosome and mitochondrial DNA must have come in through the <5% introgression from modern humans (event 3). Even though the probability of fixation for alleles received through the <5% introgression is infinitesimally small under neutrality, simulations have demonstrated that selection coefficients of >1% could make this scenario plausible. A consequence of this scenario is that we would expect to see more modern human to Neanderthal introgression on functionally constrained parts of the autosomes, however the opposite is what we infer in the real data. To address these problems, we investigate a simpler model with fewer admixture events. We hypothesize that modern humans introgressed into Neanderthals with an admixture proportion of >30%. This would accommodate the time to the common ancestor of modern humans of 410-320kya on the Y chromosome and 470-360kya on mtDNA, and be a high enough proportion of admixture so that we would not need to invoke selection to drive the modern human alleles in Neanderthals to fixation. Furthermore, we consider the possibility of no gene flow into Denisovans from an unknown archaic lineage. Instead, the previously published signal of this event could be explained as by-product of the model of deep population structure in Africa. By using high coverage whole-genome-sequence polymorphism data from 3 Neanderthals, 1 Denisovan, and numerous YRI haplotypes from the 1000 Genomes project, we use a suite of tools to investigate these models. Specifically, fastsimcoal and momi2 utilise the site-frequency-spectrum to infer the plausible evolutionary models; similarly, Relate and PSMC infer the distribution of coalescence times which is a power tool for demographic inference.       <*Standard model*>:  [A] 2-19% admixture in Africans from an “Unknown archaic” lineage which split off ~1000kya.  [B] 3-6% gene flow into Denisovans from another “Unknown archaic lineage, which split 1400-900kya.  [C] <5% gene flow into Neandertals 410-100kya, from the main modern human lineage which split off from the ancestral population of Neanderthals and Denisovans 770-550 kya.  [D] 0.5-3% gene flow into Denisovans from the main Neanderthal lineage; the main Neanderthal split from Denisovans being 470-380 kya.

11:40AM - 3) Patrick Gemmell: Scott Edwards lab

  • Title:  A phylogenetic method linking nucleotide substitution rates to continuous trait evolution
  • Summary: We present a software method that relates nucleotide substitution rates to changes in a continuous trait of interest. The method takes as input a multiple sequence alignment of conserved elements, continuous trait data observed in extant species, and a background phylogeny and substitution process. Gibbs sampling is used to assign conservation states (background, conserved, accelerated) to lineages and explore whether the assigned states are associated with increases or decreases in the rate of trait evolution. We think these sorts of methods will ultimately increase our understanding of natural history as well as shine a spotlight on parts of our own genome that are of biomedical interest.

12:00PM - Lunch 


1:00PM - 4) Paul Torrillo: Tami Lieberman lab

  • Title:  Reversions mask the contribution of adaptive evolution in microbiomes
  • Summary: Recent work has suggested that adaptive point mutations occurring within human gut microbiomes sweep through a within-host bacterial population in months. Interestingly, the same genes under adaptive evolution on this timescale appear to be under purifying selection on longer, between-host, time scales. Inspired by this discrepancy, we revisit the long-standing observation that microbial genomes show a timescale dependence of dN/dS, with purifying selection dominating on long timescales. This timescale is traditionally interpreted as a consequence of weak purifying selection slowly purging nonsynonymous mutations from the population. We show this interpretation is problematic in asexual populations; not only does this model ignore the population-wide purges of diversity that occur during adaptive sweeps, the magnitude of purifying selection required to fit observed data would imply an unrealistic rapid decay in fitness from genetic drift. Instead, we explore models in which environmental fluctuations driving adaptive mutational reversions and strong adaptive tradeoffs can obfuscate dN/dS. This work raises the possibility that interpreting low dN/dS for genes and genomes as a signal of low adaptation is misleading and many facets of evolution and ecology may currently be overlooked by relying heavily on the standard neutral model.

1:20PM - 5) Angela Early: Dan Neafsey lab

  • Title:  Selection in a shrinking malaria parasite population with low recombination
  • Summary: Drug resistance in the malaria parasite Plasmodium falciparum has recurrently evolved in geographic regions with small—not large—parasite population sizes. Effective malaria elimination therefore requires understanding the distinct evolutionary dynamics of small parasite populations. Currently, several hurdles impede the application of typical selection tests. For example, P. falciparum’s sexual, largely haploid life cycle permits long-persisting clonal lineages, skewing diversity patterns. In addition, parasite prevalence is patchy, resulting in spatial and temporal overdispersion of recombination events. New approaches are therefore needed to effectively detect selection. Here, we discuss a selection testing framework that is tailored to the high background relatedness of small, inbred parasite populations. We employ identity-by-descent analysis to define haplotype blocks, track recombination events, and identify targets of selection. We applied this approach to whole genome sequence data from over 800 P. falciparum samples collected across two decades from the Guiana Shield of South America. This parasite population recently experienced strong selection and phenotypic adaptation due to changes in drug usage, diagnostic/treatment rates, and vector habitat availability. Demographically, its transmission patterns are characterized by localized outbreaks and intermittent bursts of recombination. We have found that relatedness among Guiana Shield parasites has increased over the last two decades, reflecting a decreased disease burden and parasite population contraction. Increased relatedness is not uniform across the genome and marks likely selection on both known resistance-associated genes and novel candidate genes. Temporal analysis of the haplotypic patterns reveals changed selection signals after the introduction of a novel therapeutic. Further, the results support the hypothesis that small Plasmodium populations are not mutation-limited, as the same de novo mutation appears on distinct haplotypic backgrounds. This latter observation both impacts our understanding of parasite evolution and informs how to best monitor for and track drug resistance emergence.

1:40PM - 6) Lucas Moreira & Diane Genereux: Elinor Karlsson lab

  • Title:  Inferring the genomic basis of cellular robustness evolution in mammals
  • Summary: In humans, aberrations in body temperature, blood glucose and oxygen can damage cells, as observed in heat stroke, diabetes, and sleep apnea. But systemic homeostasis is not universal among placental mammalian species. Dromedary camels routinely reach body temperatures as high as 40°C during the hot desert day, deep-diving mammals endure transiently low blood oxygen, and fruit bats attain blood glucose levels that would damage human cells. To discover the evolutionary history of cellular robustness in these and other placental mammals, we are applying a common garden framework. We are culturing dermal fibroblasts from 12 species, including humans, across a range of temperature, glucose and oxygen conditions, recording their morphology, metabolic states, gene expression, and chromatin accessibility, then applying agent-based modeling to distinguish driver from secondary responses. Our preliminary results reveal several  pathways activated in all of our study species upon exposure to high temperature, including expression of heat shock proteins and upregulation of cholesterol biosynthesis. We have also identified responses unique to species that are not strictly homeothermic, including widespread post-transcriptional changes. Our approach holds great promise for discovering the molecular mechanisms and evolutionary history of cellular robustness across the mammalian tree.

2:00PM - 6) Yuttapong Thawornwattana: James Mallet lab

  • Title:  Inferring the direction of gene flow from genomic data
  • Summary:  Genome sequence data are informative about species divergence and gene flow.  However, opposite directions of gene flow can often lead to similar patterns in sequence data such as reduced sequence divergence between species, making inference of the direction of gene flow difficult.  Moreover, popular methods for studying gene flow often use summaries of sequence data such as genome-wide site-pattern counts.  As a result, they tend to have limited power to estimate key features of gene flow such as its direction and timing.  Here, we characterize features of genomic sequence data that are informative about the direction of gene flow in a likelihood-based multispecies-coalescent framework using a combination of mathematical analysis and computer simulation.  We find that it is easier to infer more recent gene flow than ancient one, and when there is a longer period separating the evolution between the initial divergence and subsequent gene flow.  These two factors explain why it is often easier to infer gene flow from a small population to a large one than in the opposite direction, and easier to infer inflow (gene flow from outgroup species to an ingroup species) than outflow (gene flow from an ingroup species to an outgroup species).  When gene flow is assumed to occur in the wrong direction, the timing of gene flow tends to be correctly estimated and the Bayesian test of gene flow is often significant, while the magnitude of gene flow can be overestimated.  Our findings provide useful guidance for inferring gene flow from genomic data.

2:20PM - Closing Remarks [Organizers]