Phylogeographic model selection using convolutional neural networks

Genetic structure and landscape effects on gene flow in the Neotropical lizard Norops brasiliensis (Squamata: Dactyloidae)

One key research goal of evolutionary biology is to understand the origin and maintenance of genetic variation. In the Cerrado, the South American savanna located primarily in the Central Brazilian Plateau, many hypotheses have been proposed to explain how landscape features (e.g., geographic distance, river barriers, topographic compartmentalization, and historical climatic fluctuations) have promoted genetic structure by mediating gene flow. Here, we asked whether these landscape features have influenced the genetic structure and differentiation in the lizard species Norops brasiliensis (Squamata: Dactyloidae). To achieve our goal, we used a genetic clustering analysis and estimate an effective migration surface to assess genetic structure in the focal species. Optimized isolation-by-resistance models and a simulation-based approach combined with machine learning (convolutional neural network; CNN) were then used to infer current and historical effects on population genetic structure through 12 unique landscape models. We recovered five geographically distributed populations that are separated by regions of lower-than-expected gene flow. The results of the CNN showed that geographic distance is the sole predictor of genetic variation in N. brasiliensis, and that slope, rivers, and historical climate had no discernible influence on gene flow. Our novel CNN approach was accurate (89.5%) in differentiating each landscape model. CNN and other machine learning approaches are still largely unexplored in landscape genetics studies, representing promising avenues for future research with increasingly accessible genomic datasets.

Approaches and challenges to inferring the geographical source of infectious disease outbreaks using genomic data

Genomic data hold increasing potential in the elucidation of transmission dynamics and geographical sources of infectious disease outbreaks. Phylogeographic methods that use epidemiological and genomic data obtained from surveillance enable us to infer the history of spatial transmission that is naturally embedded in the topology of phylogenetic trees as a record of the dispersal of infectious agents between geographical locations. In this Review, we provide an overview of phylogeographic approaches widely used for reconstructing the geographical sources of outbreaks of interest. These approaches can be classified into ancestral trait or state reconstruction and structured population models, with structured population models including popular structured coalescent and birth-death models. We also describe the major challenges associated with sequencing technologies, surveillance strategies, data sharing, and analysis frameworks that became apparent during the generation of large-scale genomic data in recent years, extending beyond inference approaches. Finally, we highlight the role of genomic data in geographical source inference and clarify how this enhances understanding and molecular investigations of outbreak sources.

Deep Learning in Population Genetics

Population genetics is transitioning into a data-driven discipline thanks to the availability of large-scale genomic data and the need to study increasingly complex evolutionary scenarios. With likelihood and Bayesian approaches becoming either intractable or computationally unfeasible, machine learning, and in particular deep learning, algorithms are emerging as popular techniques for population genetic inferences. These approaches rely on algorithms that learn non-linear relationships between the input data and the model parameters being estimated through representation learning from training data sets. Deep learning algorithms currently employed in the field comprise discriminative and generative models with fully connected, convolutional, or recurrent layers. Additionally, a wide range of powerful simulators to generate training data under complex scenarios are now available. The application of deep learning to empirical data sets mostly replicates previous findings of demography reconstruction and signals of natural selection in model organisms. To showcase the feasibility of deep learning to tackle new challenges, we designed a branched architecture to detect signals of recent balancing selection from temporal haplotypic data, which exhibited good predictive performance on simulated data. Investigations on the interpretability of neural networks, their robustness to uncertain training data, and creative representation of population genetic data, will provide further opportunities for technological advancements in the field.

Capturing Genetic Diversity in Seed Collections: An Empirical Study of Two Congeners with Contrasting Mating Systems

Plant mating systems shape patterns of genetic diversity and impact the long-term success of populations. As such, they are relevant to the design of seed collections aiming to maximise genetic diversity (e.g., germplasm conservation, ecological restoration). However, for most species, little is known empirically about how variation in mating systems and genetic diversity is distributed. We investigated the relationship between genetic diversity and mating systems in two functionally similar, co-occurring species of Hakea (Proteaceae), and evaluated the extent to which genetic diversity was captured in seeds. We genotyped hundreds of seedlings and mother plants via DArTseq, and developed novel implementations of two approaches to inferring the mating system from SNP data. A striking contrast in patterns of genetic diversity between H. sericea and H. teretifolia was revealed, consistent with a contrast in their mating systems. While both species had mixed mating systems, H. sericea was found to be habitually selfing, while H. teretifolia more evenly employed both selfing and outcrossing. In both species, seed collection schemes maximised genetic diversity by increasing the number of maternal lines and sites sampled, but twice as many sites were needed for the selfing species to capture equivalent levels of genetic variation at a regional scale.

dnadna: a deep learning framework for population genetics inference

MOTIVATION

We present dnadna, a flexible python-based software for deep learning inference in population genetics. It is task-agnostic and aims at facilitating the development, reproducibility, dissemination and re-usability of neural networks designed for population genetic data.

RESULTS

dnadna defines multiple user-friendly workflows. First, users can implement new architectures and tasks, while benefiting from dnadna utility functions, training procedure and test environment, which saves time and decreases the likelihood of bugs. Second, the implemented networks can be re-optimized based on user-specified training sets and/or tasks. Newly implemented architectures and pre-trained networks are easily shareable with the community for further benchmarking or other applications. Finally, users can apply pre-trained networks in order to predict evolutionary history from alternative real or simulated genetic datasets, without requiring extensive knowledge in deep learning or coding in general. dnadna comes with a peer-reviewed, exchangeable neural network, allowing demographic inference from SNP data, that can be used directly or retrained to solve other tasks. Toy networks are also available to ease the exploration of the software, and we expect that the range of available architectures will keep expanding thanks to community contributions.

AVAILABILITY AND IMPLEMENTATION

dnadna is a Python (≥3.7) package, its repository is available at gitlab.com/mlgenetics/dnadna and its associated documentation at mlgenetics.gitlab.io/dnadna/.

Congruent evolutionary responses of European steppe biota to late Quaternary climate change

Quaternary climatic oscillations had a large impact on European biogeography. Alternation of cold and warm stages caused recurrent glaciations, massive vegetation shifts, and large-scale range alterations in many species. The Eurasian steppe biome and its grasslands are a noteworthy example; they underwent climate-driven, large-scale contractions during warm stages and expansions during cold stages. Here, we evaluate the impact of these range alterations on the late Quaternary demography of several phylogenetically distant plant and insect species, typical of the Eurasian steppes. We compare three explicit demographic hypotheses by applying an approach combining convolutional neural networks with approximate Bayesian computation. We identified congruent demographic responses of cold stage expansion and warm stage contraction across all species, but also species-specific effects. The demographic history of the Eurasian steppe biota reflects major paleoecological turning points in the late Quaternary and emphasizes the role of climate as a driving force underlying patterns of genetic variance on the biome level.

Assessing model adequacy leads to more robust phylogeographic inference

Phylogeographic studies base inferences on large data sets and complex demographic models, but these models are applied in ways that could mislead researchers and compromise their inference. Researchers face three challenges associated with the use of models: (i) 'model selection', or the identification of an appropriate model for analysis; (ii) 'evaluation of analytical results', or the interpretation of the biological significance of the resulting parameter estimates, delimitations, and topologies; and (iii) 'model evaluation', or the use of statistical approaches to assess the fit of the model to the data. The field collectively invests most of its energy in point (ii) without considering the other points; we argue that attention to points (i) and (iii) is essential to phylogeographic inference.

Coalescent-based species delimitation meets deep learning: Insights from a highly fragmented cactus system

Delimiting species boundaries is a major goal in evolutionary biology. An increasing volume of literature has focused on the challenges of investigating cryptic diversity within complex evolutionary scenarios of speciation, including gene flow and demographic fluctuations. New methods based on model selection, such as approximate Bayesian computation, approximate likelihoods, and machine learning are promising tools arising in this field. Here, we introduce a framework for species delimitation using the multispecies coalescent model coupled with a deep learning algorithm based on convolutional neural networks (CNNs). We compared this strategy with a similar ABC approach. We applied both methods to test species boundary hypotheses based on current and previous taxonomic delimitations as well as genetic data (sequences from 41 loci) in Pilosocereus aurisetus, a cactus species complex with a sky-island distribution and taxonomic uncertainty. To validate our method, we also applied the same strategy on data from widely accepted species from the genus Drosophila. The results show that our CNN approach has a high capacity to distinguish among the simulated species delimitation scenarios, with higher accuracy than ABC. For the cactus data set, a splitter hypothesis without gene flow showed the highest probability in both CNN and ABC approaches, a result agreeing with previous taxonomic classifications and in line with the sky-island distribution and low dispersal of P. aurisetus. Our results highlight the cryptic diversity within the P. aurisetus complex and show that CNNs are a promising approach for distinguishing complex evolutionary histories, even outperforming the accuracy of other model-based approaches such as ABC.

The evolution of comparative phylogeography: putting the geography (and more) into comparative population genomics

Comparative population genomics is an ascendant field using genomic comparisons between species to draw inferences about forces regulating genetic variation. Comparative phylogeography, by contrast, focuses on the shared lineage histories of species codistributed geographically and is decidedly organismal in perspective. Comparative phylogeography is approximately 35 years old, and, by some metrics, is showing signs of reduced growth. Here, we contrast the goals and methods of comparative population genomics and comparative phylogeography and argue that comparative phylogeography offers an important perspective on evolutionary history that succeeds in integrating genomics with landscape evolution in ways that complement the suprageographic perspective of comparative population genomics. Focusing primarily on terrestrial vertebrates, we review the history of comparative phylogeography, its milestones and ongoing conceptual innovations, its increasingly global focus, and its status as a bridge between landscape genomics and the process of speciation. We also argue that, as a science with a strong “sense of place,” comparative phylogeography offers abundant “place-based” educational opportunities with its focus on geography and natural history, as well as opportunities for collaboration with local communities and indigenous peoples. Although comparative phylogeography does not yet require whole-genome sequencing for many of its goals, we conclude that it nonetheless plays an important role in grounding our interpretation of genetic variation in the fundamentals of geography and Earth history.