From Birdsong to Human Speech Recognition: Bayesian Inference on a Hierarchy of Nonlinear Dynamical Systems

Our knowledge about the computational mechanisms underlying human learning and recognition of sound sequences, especially speech, is still very limited. One difficulty in deciphering the exact means by which humans recognize speech is that there are scarce experimental findings at a neuronal, microscopic level. Here, we show that our neuronal-computational understanding of speech learning and recognition may be vastly improved by looking at an animal model, i.e., the songbird, which faces the same challenge as humans: to learn and decode complex auditory input, in an online fashion. Motivated by striking similarities between the human and songbird neural recognition systems at the macroscopic level, we assumed that the human brain uses the same computational principles at a microscopic level and translated a birdsong model into a novel human sound learning and recognition model with an emphasis on speech. We show that the resulting Bayesian model with a hierarchy of nonlinear dynamical systems can learn speech samples such as words rapidly and recognize them robustly, even in adverse conditions. In addition, we show that recognition can be performed even when words are spoken by different speakers and with different accents—an everyday situation in which current state-of-the-art speech recognition models often fail. The model can also be used to qualitatively explain behavioral data on human speech learning and derive predictions for future experiments.     

Mimicking human neuronal pathways in silico: an emergent model on the effective connectivity

We present a novel computational model that detects temporal configurations of a given human neuronal pathway and constructs its artificial replication. This poses a great challenge since direct recordings from individual neurons are impossible in the human central nervous system and therefore the underlying neuronal pathway has to be considered as a black box. For tackling this challenge, we used a branch of complex systems modeling called artificial self-organization in which large sets of software entities interacting locally give rise to bottom-up collective behaviors. The result is an emergent model where each software entity represents an integrate-and-fire neuron. We then applied the model to the reflex responses of single motor units obtained from conscious human subjects. Experimental results show that the model recovers functionality of real human neuronal pathways by comparing it to appropriate surrogate data. What makes the model promising is the fact that, to the best of our knowledge, it is the first realistic model to self-wire an artificial neuronal network by efficiently combining neuroscience with artificial self-organization. Although there is no evidence yet of the model’s connectivity mapping onto the human connectivity, we anticipate this model will help neuroscientists to learn much more about human neuronal networks, and could also be used for predicting hypotheses to lead future experiments.     

Reconciling emergences: An information-theoretic approach to identify causal emergence in multivariate data

The broad concept of emergence is instrumental in various of the most challenging open scientific questions—yet, few quantitative theories of what constitutes emergent phenomena have been proposed. This article introduces a formal theory of causal emergence in multivariate systems, which studies the relationship between the dynamics of parts of a system and macroscopic features of interest. Our theory provides a quantitative definition of downward causation, and introduces a complementary modality of emergent behaviour—which we refer to as causal decoupling. Moreover, the theory allows practical criteria that can be efficiently calculated in large systems, making our framework applicable in a range of scenarios of practical interest. We illustrate our findings in a number of case studies, including Conway’s Game of Life, Reynolds’ flocking model, and neural activity as measured by electrocorticography.     

Phyletic Profiling with Cliques of Orthologs Is Enhanced by Signatures of Paralogy Relationships

New microbial genomes are sequenced at a high pace, allowing insight into the genetics of not only cultured microbes, but a wide range of metagenomic collections such as the human microbiome. To understand the deluge of genomic data we face, computational approaches for gene functional annotation are invaluable. We introduce a novel model for computational annotation that refines two established concepts: annotation based on homology and annotation based on phyletic profiling. The phyletic profiling-based model that includes both inferred orthologs and paralogs—homologs separated by a speciation and a duplication event, respectively—provides more annotations at the same average Precision than the model that includes only inferred orthologs. For experimental validation, we selected 38 poorly annotated Escherichia coli genes for which the model assigned one of three GO terms with high confidence: involvement in DNA repair, protein translation, or cell wall synthesis. Results of antibiotic stress survival assays on E. coli knockout mutants showed high agreement with our model''s estimates of accuracy: out of 38 predictions obtained at the reported Precision of 60%, we confirmed 25 predictions, indicating that our confidence estimates can be used to make informed decisions on experimental validation. Our work will contribute to making experimental validation of computational predictions more approachable, both in cost and time. Our predictions for 998 prokaryotic genomes include ∼400000 specific annotations with the estimated Precision of 90%, ∼19000 of which are highly specific—e.g. “penicillin binding,” “tRNA aminoacylation for protein translation,” or “pathogenesis”—and are freely available at http://gorbi.irb.hr/.     

Migration through a Major Andean Ecogeographic Disruption as a Driver of Genetic and Phenotypic Diversity in a Wild Tomato Species

Evolutionary dynamics at the population level play a central role in creating the diversity of life on our planet. In this study, we sought to understand the origins of such population-level variation in mating systems and defensive acylsugar chemistry in Solanum habrochaites—a wild tomato species found in diverse Andean habitats in Ecuador and Peru. Using Restriction-site-Associated-DNA-Sequencing (RAD-seq) of 50 S. habrochaites accessions, we identified eight population clusters generated via isolation and hybridization dynamics of 4–6 ancestral populations. Detailed characterization of mating systems of these clusters revealed emergence of multiple self-compatible (SC) groups from progenitor self-incompatible populations in the northern part of the species range. Emergence of these SC groups was also associated with fixation of deleterious alleles inactivating acylsugar acetylation. The Amotape-Huancabamba Zone—a geographical landmark in the Andes with high endemism and isolated microhabitats—was identified as a major driver of differentiation in the northern species range, whereas large geographical distances contributed to population structure and evolution of a novel SC group in the central and southern parts of the range, where the species was also inferred to have originated. Findings presented here highlight the role of the diverse ecogeography of Peru and Ecuador in generating population differentiation, and enhance our understanding of the microevolutionary processes that create biological diversity.     

Enhanced Sensitivity to Rapid Input Fluctuations by Nonlinear Threshold Dynamics in Neocortical Pyramidal Neurons

The way in which single neurons transform input into output spike trains has fundamental consequences for network coding. Theories and modeling studies based on standard Integrate-and-Fire models implicitly assume that, in response to increasingly strong inputs, neurons modify their coding strategy by progressively reducing their selective sensitivity to rapid input fluctuations. Combining mathematical modeling with in vitro experiments, we demonstrate that, in L5 pyramidal neurons, the firing threshold dynamics adaptively adjust the effective timescale of somatic integration in order to preserve sensitivity to rapid signals over a broad range of input statistics. For that, a new Generalized Integrate-and-Fire model featuring nonlinear firing threshold dynamics and conductance-based adaptation is introduced that outperforms state-of-the-art neuron models in predicting the spiking activity of neurons responding to a variety of in vivo-like fluctuating currents. Our model allows for efficient parameter extraction and can be analytically mapped to a Generalized Linear Model in which both the input filter—describing somatic integration—and the spike-history filter—accounting for spike-frequency adaptation—dynamically adapt to the input statistics, as experimentally observed. Overall, our results provide new insights on the computational role of different biophysical processes known to underlie adaptive coding in single neurons and support previous theoretical findings indicating that the nonlinear dynamics of the firing threshold due to Na⁺-channel inactivation regulate the sensitivity to rapid input fluctuations.     

Identification of emergent motion compartments in the amniote embryo

The tissue scale deformations (≥1mm) required to form an amniote embryo are poorly understood. Here, we studied ∼400 μm-sized explant units from gastrulating quail embryos. The explants deformed in a reproducible manner when grown using a novel vitelline membrane-based culture method. Time-lapse recordings of latent embryonic motion patterns were analyzed after disk-shaped tissue explants were excised from three specific regions near the primitive streak: 1) anterolateral epiblast, 2) posterolateral epiblast, and 3) the avian organizer (Hensen''s node). The explants were cultured for 8 hours—an interval equivalent to gastrulation. Both the anterolateral and the posterolateral epiblastic explants engaged in concentric radial/centrifugal tissue expansion. In sharp contrast, Hensen''s node explants displayed Cartesian-like, elongated, bipolar deformations—a pattern reminiscent of axis elongation. Time-lapse analysis of explant tissue motion patterns indicated that both cellular motility and extracellular matrix fiber (tissue) remodeling take place during the observed morphogenetic deformations. As expected, treatment of tissue explants with a selective Rho-Kinase (p160ROCK) signaling inhibitor, Y27632, completely arrested all morphogenetic movements. Microsurgical experiments revealed that lateral epiblastic tissue was dispensable for the generation of an elongated midline axis— provided that an intact organizer (node) is present. Our computational analyses suggest the possibility of delineating tissue-scale morphogenetic movements at anatomically discrete locations in the embryo. Further, tissue deformation patterns, as well as the mechanical state of the tissue, require normal actomyosin function. We conclude that amniote embryos contain tissue-scale, regionalized morphogenetic motion generators, which can be assessed using our novel computational time-lapse imaging approach. These data and future studies—using explants excised from overlapping anatomical positions—will contribute to understanding the emergent tissue flow that shapes the amniote embryo.     

Harnessing natural diversity to probe metabolic pathways

Analyses of cellular processes in the yeast Saccharomyces cerevisiae rely primarily upon a small number of highly domesticated laboratory strains, leaving the extensive natural genetic diversity of the model organism largely unexplored and unexploited. We asked if this diversity could be used to enrich our understanding of basic biological processes. As a test case, we examined a simple trait: the utilization of di/tripeptides as nitrogen sources. The capacity to import small peptides is likely to be under opposing selective pressures (nutrient utilization versus toxin vulnerability) and may therefore be sculpted by diverse pathways and strategies. Hitherto, dipeptide utilization in S. cerevisiae was solely ascribed to the activity of a single protein, the Ptr2p transporter. Using high-throughput phenotyping and several genetically diverse strains, we identified previously unknown cellular activities that contribute to this trait. We find that the Dal5p allantoate/ureidosuccinate permease is also capable of facilitating di/tripeptide transport. Moreover, even in the absence of Dal5p and Ptr2p, an additional activity—almost certainly the periplasmic asparaginase II Asp3p—facilitates the utilization of dipeptides with C-terminal asparagine residues by a different strategy. Another, as-yet-unidentified activity enables the utilization of dipeptides with C-terminal arginine residues. The relative contributions of these activities to the utilization of di/tripeptides vary among the strains analyzed, as does the vulnerability of these strains to a toxic dipeptide. Only by sampling the genetic diversity of multiple strains were we able to uncover several previously unrecognized layers of complexity in this metabolic pathway. High-throughput phenotyping facilitates the rapid exploration of the molecular basis of biological complexity, allowing for future detailed investigation of the selective pressures that drive microbial evolution.     

SVM2Motif—Reconstructing Overlapping DNA Sequence Motifs by Mimicking an SVM Predictor

Identifying discriminative motifs underlying the functionality and evolution of organisms is a major challenge in computational biology. Machine learning approaches such as support vector machines (SVMs) achieve state-of-the-art performances in genomic discrimination tasks, but—due to its black-box character—motifs underlying its decision function are largely unknown. As a remedy, positional oligomer importance matrices (POIMs) allow us to visualize the significance of position-specific subsequences. Although being a major step towards the explanation of trained SVM models, they suffer from the fact that their size grows exponentially in the length of the motif, which renders their manual inspection feasible only for comparably small motif sizes, typically k ≤ 5. In this work, we extend the work on positional oligomer importance matrices, by presenting a new machine-learning methodology, entitled motifPOIM, to extract the truly relevant motifs—regardless of their length and complexity—underlying the predictions of a trained SVM model. Our framework thereby considers the motifs as free parameters in a probabilistic model, a task which can be phrased as a non-convex optimization problem. The exponential dependence of the POIM size on the oligomer length poses a major numerical challenge, which we address by an efficient optimization framework that allows us to find possibly overlapping motifs consisting of up to hundreds of nucleotides. We demonstrate the efficacy of our approach on a synthetic data set as well as a real-world human splice site data set.     

Where Next for Microbiome Research?

The development of high-throughput sequencing technologies has transformed our capacity to investigate the composition and dynamics of the microbial communities that populate diverse habitats. Over the past decade, these advances have yielded an avalanche of metagenomic data. The current stage of “van Leeuwenhoek”–like cataloguing, as well as functional analyses, will likely accelerate as DNA and RNA sequencing, plus protein and metabolic profiling capacities and computational tools, continue to improve. However, it is time to consider: what’s next for microbiome research? The short pieces included here briefly consider the challenges and opportunities awaiting microbiome research.

This Perspective is part of the “Where next?” Series.

Soon, we will enter an era when “the number of population genomes deposited in public databases will dwarf those from isolates and single cells” (Gene Tyson). Clearly, as all authors noted in the following, our focus will move from describing the composition of microbial communities to elucidating the principles that govern their assembly, dynamics, and functions. How will such principles be discovered? Elhanan Borenstein proposes that a systems biology–based approach, particularly the development of mathematical and computational models of the interactions between the specific community components, will be critical for understanding the function and dynamics of microbiomes. Evolutionary biologists Howard Ochman and Andrew Moeller want to decipher how microbial assemblies evolve but challenge us to also consider the role of microbial communities in organismal evolution, and they make the exciting prediction that microbes will be implicated in the evolution of eusociality and cooperation. Brett Finlay underscores the need for deciphering the mechanistic bases—particularly the chemical/metabolite signals—for interactions between members of microbial communities and their hosts. He emphasizes how this knowledge will enable creation of new tools to manipulate the microbiota, a key challenge for future investigation. Heidi Kong also encourages deciphering the mechanisms that underlie associations between particular skin surfaces and disorders and their respective microbiota. Jeffrey Gordon considers several intriguing opportunities as well as challenges that manipulation of the gut microbiota presents for improved human nutrition and health. Finally, Karen Nelson, Karim Dabbagh and Hamilton Smith suggest that using synthetic genomes to create novel microbes or even synthetic microbiomes offers a new way to engineer the microbiota. Overall, future microbiome research regarding the molecules and mechanisms mediating interactions between members of microbial communities and their hosts should lead to discovery of exciting new biology and transformative therapeutics.     

The Evolutionary Origins of Hierarchy

Hierarchical organization—the recursive composition of sub-modules—is ubiquitous in biological networks, including neural, metabolic, ecological, and genetic regulatory networks, and in human-made systems, such as large organizations and the Internet. To date, most research on hierarchy in networks has been limited to quantifying this property. However, an open, important question in evolutionary biology is why hierarchical organization evolves in the first place. It has recently been shown that modularity evolves because of the presence of a cost for network connections. Here we investigate whether such connection costs also tend to cause a hierarchical organization of such modules. In computational simulations, we find that networks without a connection cost do not evolve to be hierarchical, even when the task has a hierarchical structure. However, with a connection cost, networks evolve to be both modular and hierarchical, and these networks exhibit higher overall performance and evolvability (i.e. faster adaptation to new environments). Additional analyses confirm that hierarchy independently improves adaptability after controlling for modularity. Overall, our results suggest that the same force–the cost of connections–promotes the evolution of both hierarchy and modularity, and that these properties are important drivers of network performance and adaptability. In addition to shedding light on the emergence of hierarchy across the many domains in which it appears, these findings will also accelerate future research into evolving more complex, intelligent computational brains in the fields of artificial intelligence and robotics.     

Cerebral hierarchies: predictive processing,precision and the pulvinar

This paper considers neuronal architectures from a computational perspective and asks what aspects of neuroanatomy and neurophysiology can be disclosed by the nature of neuronal computations? In particular, we extend current formulations of the brain as an organ of inference—based upon hierarchical predictive coding—and consider how these inferences are orchestrated. In other words, what would the brain require to dynamically coordinate and contextualize its message passing to optimize its computational goals? The answer that emerges rests on the delicate (modulatory) gain control of neuronal populations that select and coordinate (prediction error) signals that ascend cortical hierarchies. This is important because it speaks to a hierarchical anatomy of extrinsic (between region) connections that form two distinct classes, namely a class of driving (first-order) connections that are concerned with encoding the content of neuronal representations and a class of modulatory (second-order) connections that establish context—in the form of the salience or precision ascribed to content. We explore the implications of this distinction from a formal perspective (using simulations of feature–ground segregation) and consider the neurobiological substrates of the ensuing precision-engineered dynamics, with a special focus on the pulvinar and attention.     

Large scale analysis of the mutational landscape in HT-SELEX improves aptamer discovery

High-Throughput (HT) SELEX combines SELEX (Systematic Evolution of Ligands by EXponential Enrichment), a method for aptamer discovery, with massively parallel sequencing technologies. This emerging technology provides data for a global analysis of the selection process and for simultaneous discovery of a large number of candidates but currently lacks dedicated computational approaches for their analysis. To close this gap, we developed novel in-silico methods to analyze HT-SELEX data and utilized them to study the emergence of polymerase errors during HT-SELEX. Rather than considering these errors as a nuisance, we demonstrated their utility for guiding aptamer discovery. Our approach builds on two main advancements in aptamer analysis: AptaMut—a novel technique allowing for the identification of polymerase errors conferring an improved binding affinity relative to the ‘parent’ sequence and AptaCluster—an aptamer clustering algorithm which is to our best knowledge, the only currently available tool capable of efficiently clustering entire aptamer pools. We applied these methods to an HT-SELEX experiment developing aptamers against Interleukin 10 receptor alpha chain (IL-10RA) and experimentally confirmed our predictions thus validating our computational methods.     

Use of artificial genomes in assessing methods for atypical gene detection

Parametric methods for identifying laterally transferred genes exploit the directional mutational biases unique to each genome. Yet the development of new, more robust methods—as well as the evaluation and proper implementation of existing methods—relies on an arbitrary assessment of performance using real genomes, where the evolutionary histories of genes are not known. We have used the framework of a generalized hidden Markov model to create artificial genomes modeled after genuine genomes. To model a genome, “core” genes—those displaying patterns of mutational biases shared among large numbers of genes—are identified by a novel gene clustering approach based on the Akaike information criterion. Gene models derived from multiple “core” gene clusters are used to generate an artificial genome that models the properties of a genuine genome. Chimeric artificial genomes—representing those having experienced lateral gene transfer—were created by combining genes from multiple artificial genomes, and the performance of the parametric methods for identifying “atypical” genes was assessed directly. We found that a hidden Markov model that included multiple gene models, each trained on sets of genes representing the range of genotypic variability within a genome, could produce artificial genomes that mimicked the properties of genuine genomes. Moreover, different methods for detecting foreign genes performed differently—i.e., they had different sets of strengths and weaknesses—when identifying atypical genes within chimeric artificial genomes.     

Resource requirements for ecosystem conservation: A combined industrial and natural ecology approach to quantifying natural capital use in nature

Socioeconomic demand for natural capital is causing catastrophic losses of biodiversity and ecosystem functionality, most notably in regions where socioeconomic‐and eco‐systems compete for natural capital, e.g., energy (animal or plant matter). However, a poor quantitative understanding of what natural capital is needed to support biodiversity in ecosystems, while at the same time satisfy human development needs—those associated with human development within socioeconomic systems—undermines our ability to sustainably manage global stocks of natural capital. Here we describe a novel concept and accompanying methodology (relating the adult body mass of terrestrial species to their requirements for land area, water, and energy) to quantify the natural capital needed to support terrestrial species within ecosystems, analogous to how natural capital use by humans is quantified in a socioeconomic context. We apply this methodology to quantify the amount of natural capital needed to support species observed using a specific surveyed site in Scotland. We find that the site can support a larger assemblage of species than those observed using the site; a primary aim of the rewilding project taking place there. This method conceptualises, for the first time, a comprehensive “dual‐system” approach: modelling natural capital use in socioeconomic‐and eco‐systems simultaneously. It can facilitate the management of natural capital at the global scale, and in both the conservation and creation (e.g., rewilding) of biodiversity within managed ecosystems, representing an advancement in determining what socioeconomic trade‐offs are needed to achieve contemporary conservation targets alongside ongoing human development.     

A scenario-based approach to modeling development: a prototype model of C. elegans vulval fate specification

Studies of developmental biology are often facilitated by diagram “models” that summarize the current understanding of underlying mechanisms. The increasing complexity of our understanding of development necessitates computational models that can extend these representations to include their dynamic behavior. Here we present a prototype model of Caenorhabditis elegans vulval precursor cell fate specification that represents many processes crucial for this developmental event but that are hard to integrate using other modeling methodologies. We demonstrate the integrative capabilities of our methodology by comprehensively incorporating the contents of three seminal papers, showing that this methodology can lead to comprehensive models of developmental biology. The prototype computational model was built and is run using a language (Live Sequence Charts) and tool (the Play-Engine) that facilitate the same conceptual processes biologists use to construct and probe diagram-type models. We demonstrate that this modeling approach permits rigorous tests of mutual consistency between experimental data and mechanistic hypotheses and can identify specific conflicting results, providing a useful approach to probe developmental systems.     

Mechanistic model of nutrient uptake explains dichotomy between marine oligotrophic and copiotrophic bacteria

Marine bacterial diversity is immense and believed to be driven in part by trade-offs in metabolic strategies. Here we consider heterotrophs that rely on organic carbon as an energy source and present a molecular-level model of cell metabolism that explains the dichotomy between copiotrophs—which dominate in carbon-rich environments—and oligotrophs—which dominate in carbon-poor environments—as the consequence of trade-offs between nutrient transport systems. While prototypical copiotrophs, like Vibrios, possess numerous phosphotransferase systems (PTS), prototypical oligotrophs, such as SAR11, lack PTS and rely on ATP-binding cassette (ABC) transporters, which use binding proteins. We develop models of both transport systems and use them in proteome allocation problems to predict the optimal nutrient uptake and metabolic strategy as a function of carbon availability. We derive a Michaelis–Menten approximation of ABC transport, analytically demonstrating how the half-saturation concentration is a function of binding protein abundance. We predict that oligotrophs can attain nanomolar half-saturation concentrations using binding proteins with only micromolar dissociation constants and while closely matching transport and metabolic capacities. However, our model predicts that this requires large periplasms and that the slow diffusion of the binding proteins limits uptake. Thus, binding proteins are critical for oligotrophic survival yet severely constrain growth rates. We propose that this trade-off fundamentally shaped the divergent evolution of oligotrophs and copiotrophs.     

Single nucleotide polymorphism arrays: a decade of biological, computational and technological advances

Array manufacturers originally designed single nucleotide polymorphism (SNP) arrays to genotype human DNA at thousands of SNPs across the genome simultaneously. In the decade since their initial development, the platform's applications have expanded to include the detection and characterization of copy number variation—whether somatic, inherited, or de novo—as well as loss-of-heterozygosity in cancer cells. The technology's impressive contributions to insights in population and molecular genetics have been fueled by advances in computational methodology, and indeed these insights and methodologies have spurred developments in the arrays themselves. This review describes the most commonly used SNP array platforms, surveys the computational methodologies used to convert the raw data into inferences at the DNA level, and details the broad range of applications. Although the long-term future of SNP arrays is unclear, cost considerations ensure their relevance for at least the next several years. Even as emerging technologies seem poised to take over for at least some applications, researchers working with these new sources of data are adopting the computational approaches originally developed for SNP arrays.