Fitness Landscape Research Articles

The dynamics of evolutionary branching in an ecological model

Eco-evolutionary modelling involves the coupling of ecological equations to evolutionary ones. The interaction between ecological dynamics and evolutionary processes is essential to simulating evolutionary branching, a precursor to speciation. The creation and maintenance of biodiversity in models depends upon their ability to capture the dynamics of evolutionary branching. Understanding these systems requires low-dimension models that are amenable to analysis. The rapid reproduction rates of marine plankton ecosystems and their importance in determining the fluxes of climatically important gases between the ocean and atmosphere suggest that the next generation of global climate models needs to incorporate eco-evolutionary models in the ocean. This requires simple population-level models, that can represent such eco-evolutionary processes with orders of magnitude fewer equations than models that follow the dynamics of individual phenotypes. We present a general framework for developing eco-evolutionary models and consider its general properties. This framework defines a fitness function and assumes a beta distribution of phenotype abundances within each population. It simulates the change in total population size, the mean trait value, and the trait differentiation, from which the variance of trait values in the population may be calculated. We test the efficacy of the eco-evolutionary modelling framework by comparing the dynamics of evolutionary branching in a six-equation eco-evolutionary model that has evolutionary branching, with that of an equivalent one-hundred equation model that simulates the dynamics of every phenotype in the population. The latter model does not involve a population fitness function, nor does it assume a distribution of phenotype abundance across trait values. The eco-evolutionary population model and the phenotype model produce similar evolutionary branching, both qualitatively and quantitatively, in both symmetric and asymmetric fitness landscapes. In order to better understand the six-equation model, we develop a heuristic three-equation eco-evolutionary model. We use the density-independent mortality parameter as a convenient bifurcation parameter, so that differences in evolutionary branching dynamics in symmetric and asymmetric fitness landscapes may be investigated. This model shows that evolutionary branching of a stable population is flagged by a zero in the local trait curvature; the trait curvature then changes sign from negative to positive and back to negative, along the solution. It suggests that evolutionary branching points may be generated differently, with different dynamical properties, depending upon, in this case, the symmetry of the system. It also suggests that a changing environment, that may change attributes such as mortality, could have profound effects on an ecosystem’s ability to adapt. Our results suggest that the properties of the three-dimensional model can provide useful insights into the properties of the higher-dimension models. In particular, the bifurcation properties of the simple model predict the processes by which the more complicated models produce evolutionary branching points. The corresponding bifurcation properties of the phenotype and population models, evident in the dynamics of the phenotype distributions they predict, suggest that our eco-evolutionary modelling framework captures the essential properties that underlie the evolution of phenotypes in populations.

How to Sample Dozens of Substitutions per Site with λ Dynamics.

Alchemical free energy methods are useful in computer-aided drug design and computational protein design because they provide rigorous statistical mechanics-based estimates of free energy differences from molecular dynamics simulations. λ dynamics is a free energy method with the ability to characterize combinatorial chemical spaces spanning thousands of related systems within a single simulation, which gives it a distinct advantage over other alchemical free energy methods that are mostly limited to pairwise comparisons. Recently developed methods have improved the scalability of λ dynamics to perturbations at many sites; however, the size of chemical space that can be explored at each individual site has previously been limited to fewer than ten substituents. As the number of substituents increases, the volume of alchemical space corresponding to nonphysical alchemical intermediates grows exponentially relative to the size corresponding to the physical states of interest. Beyond nine substituents, λ dynamics simulations become lost in an alchemical morass of intermediate states. In this work, we introduce new biasing potentials that circumvent excessive sampling of intermediate states by favoring sampling of physical end points relative to alchemical intermediates. Additionally, we present a more scalable adaptive landscape flattening algorithm for these larger alchemical spaces. Finally, we show that this potential enables more efficient sampling in both protein and drug design test systems with up to 24 substituents per site, enabling, for the first time, simultaneous simulation of all 20 amino acids.

A Delayed Spiking Neural Membrane System for Adaptive Nearest Neighbor-Based Density Peak Clustering.

Although the density peak clustering (DPC) algorithm can effectively distribute samples and quickly identify noise points, it lacks adaptability and cannot consider the local data structure. In addition, clustering algorithms generally suffer from high time complexity. Prior research suggests that clustering algorithms grounded in P systems can mitigate time complexity concerns. Within the realm of membrane systems (P systems), spiking neural P systems (SN P systems), inspired by biological nervous systems, are third-generation neural networks that possess intricate structures and offer substantial parallelism advantages. Thus, this study first improved the DPC by introducing the maximum nearest neighbor distance and K-nearest neighbors (KNN). Moreover, a method based on delayed spiking neural P systems (DSN P systems) was proposed to improve the performance of the algorithm. Subsequently, the DSNP-ANDPC algorithm was proposed. The effectiveness of DSNP-ANDPC was evaluated through comprehensive evaluations across four synthetic datasets and 10 real-world datasets. The proposed method outperformed the other comparison methods in most cases.

Stagnation Detection in Highly Multimodal Fitness Landscapes

Stagnation detection has been proposed as a mechanism for randomized search heuristics to escape from local optima by automatically increasing the size of the neighborhood to find the so-called gap size, i. e., the distance to the next improvement. Its usefulness has mostly been considered in simple multimodal landscapes with few local optima that could be crossed one after another. In multimodal landscapes with a more complex location of optima of similar gap size, stagnation detection suffers from the fact that the neighborhood size is frequently reset to 1 without using gap sizes that were promising in the past. In this paper, we investigate a new mechanism called radius memory which can be added to stagnation detection to control the search radius more carefully by giving preference to values that were successful in the past. We implement this idea in an algorithm called SD-RLSm\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^{\ ext {m}}$$\\end{document} and show compared to previous variants of stagnation detection that it yields speed-ups for linear functions under uniform constraints and the minimum spanning tree problem. Moreover, its running time does not significantly deteriorate on unimodal functions and a generalization of the Jump benchmark. Finally, we present experimental results carried out to study SD-RLSm\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^{\ ext {m}}$$\\end{document} and compare it with other algorithms.

Enhancing efficiency of protein language models with minimal wet-lab data through few-shot learning

Accurately modeling the protein fitness landscapes holds great importance for protein engineering. Pre-trained protein language models have achieved state-of-the-art performance in predicting protein fitness without wet-lab experimental data, but their accuracy and interpretability remain limited. On the other hand, traditional supervised deep learning models require abundant labeled training examples for performance improvements, posing a practical barrier. In this work, we introduce FSFP, a training strategy that can effectively optimize protein language models under extreme data scarcity for fitness prediction. By combining meta-transfer learning, learning to rank, and parameter-efficient fine-tuning, FSFP can significantly boost the performance of various protein language models using merely tens of labeled single-site mutants from the target protein. In silico benchmarks across 87 deep mutational scanning datasets demonstrate FSFP’s superiority over both unsupervised and supervised baselines. Furthermore, we successfully apply FSFP to engineer the Phi29 DNA polymerase through wet-lab experiments, achieving a 25% increase in the positive rate. These results underscore the potential of our approach in aiding AI-guided protein engineering.

The role of epistasis in evolutionary rescue.

The process by which adaptive evolution preserves a population threatened with extinction due to environmental changes is known as evolutionary rescue. Several factors determine the fate of those populations, including demography and genetic factors, such as standing genetic variation, gene flow, availability of de novo mutations, and so on. Despite the extensive debate about evolutionary rescue in the current literature, a study about the role of epistasis and the topography of the fitness landscape on the fate of dwindling populations is missing. In the current work, we aim to fill this gap and study the influence of epistasis on the probability of extinction of populations. We present simulation results, and analytical approximations are derived. Counterintuitively, we show that the likelihood of extinction is smaller when the degree of epistasis is higher. The reason underneath is twofold: first, higher epistasis can promote mutations of more significant phenotypic effects, but also, the incongruence between the maps genotype-phenotype and phenotype-fitness turns the fitness landscape at low epistasis more rugged, thus curbing some of its advantages.

Diving into broad-scale and high-resolution population genomics to decipher drivers of structure and climatic vulnerability in a marine invertebrate.

Species with widespread distributions play a crucial role in our understanding of climate change impacts on population structure. In marine species, population structure is often governed by both high connectivity potential and selection across strong environmental gradients. Despite the complexity of factors influencing marine populations, studying species with broad distribution can provide valuable insights into the relative importance of these factors and the consequences of climate-induced alterations across environmental gradients. We used the northern shrimp Pandalus borealis and its wide latitudinal distribution to identify current drivers of population structure and predict the species' vulnerability to climate change. A total of 1514 individuals sampled across 24° latitude were genotyped at high geographic (54 stations) and genetic (14,331 SNPs) resolutions to assess genetic variation and environmental correlations. Four populations were identified in addition to finer substructure associated with local adaptation. Geographic patterns of neutral population structure reflected predominant oceanographic currents, while a significant proportion of the genetic variation was associated with gradients in salinity and temperature. Adaptive landscapes generated using climate projections suggest a larger genomic offset in the southern extent of the P. borealis range, where shrimp had the largest adaptive standing genetic variation. Our genomic results combined with recent observations point to further deterioration in southern regions and an impending vulnerable status in the regions at higher latitudes for P. borealis. They also provide rare insights into the drivers of population structure and climatic vulnerability of a widespread meroplanktonic species, which is crucial to understanding future challenges associated with invertebrates essential to ecosystem functioning.

Improved understanding of calibration efficiency, difficulty and parameter uniqueness of conceptual rainfall runoff models using fitness landscape metrics

The ease and efficiency with which conceptual rainfall runoff (CRR) models can be calibrated, as well as issues related to the uniqueness of their parameters, has received significant attention in literature. While several studies have tried to gain a better understanding of the underlying factors affecting these issues by examining the features of model error surfaces, this has generally been done in an ad-hoc fashion using lower-dimensional representations of higher-dimensional surfaces. In this paper, it is suggested that exploratory landscape analysis (ELA) metrics can be used to quantify key features of the error surfaces of CRR models, including their roughness and flatness, as well as their degree of optima dispersion throughout the surface. This enables key error surface features of CRR models to be compared in a consistent, efficient and easily communicable fashion for models with different combinations of attributes (e.g. model structure, catchment climate conditions, error metrics, and calibration data set lengths). Results from the application of ELA metrics to the error surfaces of 420 CRR models with different combinations of the above attributes indicate that increasing model complexity results in an increase in relative error surface roughness and relative optima dispersion and that, while increasing catchment wetness increases the relative roughness of error surfaces, it also decreases optima dispersion. This suggests that for the models considered in this study, optimisation efficiency is likely to decrease with increasing model complexity and catchment wetness, while optimisation difficulty is likely to increase and parameter uniqueness likely to decrease with model complexity and catchment dryness. While implications for choice of model complexity will need further work, this study highlights the potential value of the proposed approach to understanding the calibration efficiency, difficulty and parameter uniqueness of conceptual rainfall runoff models.

Multistate Gene Cluster Switches Determine the Adaptive Mitochondrial and Metabolic Landscape of Breast Cancer.

Adaptive metabolic switches are proposed to underlie conversions between cellular states during normal development as well as in cancer evolution. Metabolic adaptations represent important therapeutic targets in tumors, highlighting the need to characterize the full spectrum, characteristics, and regulation of the metabolic switches. To investigate the hypothesis that metabolic switches associated with specific metabolic states can be recognized by locating large alternating gene expression patterns, we developed a method to identify interspersed gene sets by massive correlated biclustering and to predict their metabolic wiring. Testing the method on breast cancer transcriptome datasets revealed a series of gene sets with switch-like behavior that could be used to predict mitochondrial content, metabolic activity, and central carbon flux in tumors. The predictions were experimentally validated by bioenergetic profiling and metabolic flux analysis of 13C-labeled substrates. The metabolic switch positions also distinguished between cellular states, correlating with tumor pathology, prognosis, and chemosensitivity. The method is applicable to any large and heterogeneous transcriptome dataset to discover metabolic and associated pathophysiological states. Significance: A method for identifying the transcriptomic signatures of metabolic switches underlying divergent routes of cellular transformation stratifies breast cancer into metabolic subtypes, predicting their biology, architecture, and clinical outcome.

Fitness landscape of substrate-adaptive mutations in evolved amino acid-polyamine-organocation transporters.

The emergence of new protein functions is crucial for the evolution of organisms. This process has been extensively researched for soluble enzymes, but it is largely unexplored for membrane transporters, even though the ability to acquire new nutrients from a changing environment requires evolvability of transport functions. Here, we demonstrate the importance of environmental pressure in obtaining a new activity or altering a promiscuous activity in members of the amino acid-polyamine-organocation (APC)-type yeast amino acid transporters family. We identify APC members that have broader substrate spectra than previously described. Using in vivo experimental evolution, we evolve two of these transporter genes, AGP1 and PUT4, toward new substrate specificities. Single mutations on these transporters are found to be sufficient for expanding the substrate range of the proteins, while retaining the capacity to transport all original substrates. Nonetheless, each adaptive mutation comes with a distinct effect on the fitness for each of the original substrates, illustrating a trade-off between the ancestral and evolved functions. Collectively, our findings reveal how substrate-adaptive mutations in membrane transporters contribute to fitness and provide insights into how organisms can use transporter evolution to explore new ecological niches.

Gauge fixing for sequence-function relationships.

Quantitative models of sequence-function relationships are ubiquitous in computational biology, e.g., for modeling the DNA binding of transcription factors or the fitness landscapes of proteins. Interpreting these models, however, is complicated by the fact that the values of model parameters can often be changed without affecting model predictions. Before the values of model parameters can be meaningfully interpreted, one must remove these degrees of freedom (called "gauge freedoms" in physics) by imposing additional constraints (a process called "fixing the gauge"). However, strategies for fixing the gauge of sequence-function relationships have received little attention. Here we derive an analytically tractable family of gauges for a large class of sequence-function relationships. These gauges are derived in the context of models with all-order interactions, but an important subset of these gauges can be applied to diverse types of models, including additive models, pairwise-interaction models, and models with higher-order interactions. Many commonly used gauges are special cases of gauges within this family. We demonstrate the utility of this family of gauges by showing how different choices of gauge can be used both to explore complex activity landscapes and to reveal simplified models that are approximately correct within localized regions of sequence space. The results provide practical gauge-fixing strategies and demonstrate the utility of gauge-fixing for model exploration and interpretation.

Natural variation in neuraminidase activity influences the evolutionary potential of the seasonal H1N1 lineage hemagglutinin.

The antigenic evolution of the influenza A virus hemagglutinin (HA) gene poses a major challenge for the development of vaccines capable of eliciting long-term protection. Prior efforts to understand the mechanisms that govern viral antigenic evolution mainly focus on HA in isolation, ignoring the fact that HA must act in concert with the viral neuraminidase (NA) during replication and spread. Numerous studies have demonstrated that the degree to which the receptor-binding avidity of HA and receptor-cleaving activity of NA are balanced with each other influences overall viral fitness. We recently showed that changes in NA activity can significantly alter the mutational fitness landscape of HA in the context of a lab-adapted virus strain. Here, we test whether natural variation in relative NA activity can influence the evolutionary potential of HA in the context of the seasonal H1N1 lineage (pdmH1N1) that has circulated in humans since the 2009 pandemic. We observed substantial variation in the relative activities of NA proteins encoded by a panel of H1N1 vaccine strains isolated between 2009 and 2019. We comprehensively assessed the effect of NA background on the HA mutational fitness landscape in the circulating pdmH1N1 lineage using deep mutational scanning and observed pronounced epistasis between NA and residues in or near the receptor-binding site of HA. To determine whether NA variation could influence the antigenic evolution of HA, we performed neutralizing antibody selection experiments using a panel of monoclonal antibodies targeting different HA epitopes. We found that the specific antibody escape profiles of HA were highly contingent upon NA background. Overall, our results indicate that natural variation in NA activity plays a significant role in governing the evolutionary potential of HA in the currently circulating pdmH1N1 lineage.

Visual analysis of fitness landscapes in architectural design optimization

In architectural design optimization, fitness landscapes are used to visualize design space parameters in relation to one or more objective functions for which they are being optimized. In our design study with domain experts, we developed a visual analytics framework for exploring and analyzing fitness landscapes spanning data, projection, and visualization layers. Within the data layer, we employ two surrogate models and three sampling strategies to efficiently generate a wide array of landscapes. On the projection layer, we use star coordinates and UMAP as two alternative methods for obtaining a 2D embedding of the design space. Our interactive user interface can visualize fitness landscapes as a continuous density map or a discrete glyph-based map. We investigate the influence of surrogate models and sampling strategies on the resulting fitness landscapes in a parameter study. Additionally, we present findings from a user study (N = 12), revealing how experts’ preferences regarding projection methods and visual representations may be influenced by their level of expertise, characteristics of the techniques, and the specific task at hand. Furthermore, we demonstrate the usability and usefulness of our framework by a case study from the architecture domain, involving one domain expert.

Sampling Strategies for Experimentally Mapping Molecular Fitness Landscapes Using High-Throughput Methods.

Empirical studies of genotype-phenotype-fitness maps of proteins are fundamental to understanding the evolutionary process, in elucidating the space of possible genotypes accessible through mutations in a landscape of phenotypes and fitness effects. Yet, comprehensively mapping molecular fitness landscapes remains challenging since all possible combinations of amino acid substitutions for even a few protein sites are encoded by an enormous genotype space. High-throughput mapping of genotype space can be achieved using large-scale screening experiments known as multiplexed assays of variant effect (MAVEs). However, to accommodate such multi-mutational studies, the size of MAVEs has grown to the point where a priori determination of sampling requirements is needed. To address this problem, we propose calculations and simulation methods to approximate minimum sampling requirements for multi-mutational MAVEs, which we combine with a new library construction protocol to experimentally validate our approximation approaches. Analysis of our simulated data reveals how sampling trajectories differ between simulations of nucleotide versus amino acid variants and among mutagenesis schemes. For this, we show quantitatively that marginal gains in sampling efficiency demand increasingly greater sampling effort when sampling for nucleotide sequences over their encoded amino acid equivalents. We present a new library construction protocol that efficiently maximizes sequence variation, and demonstrate using ultradeep sequencing that the library encodes virtually all possible combinations of mutations within the experimental design. Insights learned from our analyses together with the methodological advances reported herein are immediately applicable toward pooled experimental screens of arbitrary design, enabling further assay upscaling and expanded testing of genotype space.

Initiating and integrating user-centred design in upgrading the e-learning layout: a case study on FlexLearn (Brightspace) 2.0 template

PurposeThe purpose of this paper is to articulate how the user experience (UX) approach was initiated and integrated into the centre’s scope of operations with the objective of improving the e-learning layout on the D2L learning management system (LMS). One of the most effective ways to collect user feedback has historically been to evaluate user interfaces using strategies from user testing. The integration of a UX approach by the Centre for ODL Experiences (COLE) at Wawasan Open University has led to a more user-oriented design of FlexLearn by conducting user testing on students as the target users of the platform and gathering course leaders’ (CLs) feedback after the presentation of the new template.Design/methodology/approachSince the process of design and development is a looping process, the first user testing methods employed were observation and interviews, which were conducted over the course of numerous sessions. The data collection used a mixed-methods approach, combining quantitative demographic and background data with qualitative feedback from open-ended questions and real-time interview responses. A standardized questionnaire gathered demographic information, while questions for feedback forms and interviews were adjusted based on specific tasks to explore usability and user interactions comprehensively.FindingsThe findings revealed overall positive feedback, with some concerns highlighted by the students who claimed to have trouble navigating the courses during the initial prototype. In addition to the qualitative data from the user testing session, a quantitative method based on an online questionnaire was also utilised for the CLs after the presentation of the final layout. Positive responses were received from the CLs, and constructive suggestions were considered for FlexLearn 3.0.Research limitations/implicationsThis paper is among the first that articulates the process of initiating and integrating user-centred design in an effort to improve the user experience of online and ODL platforms and LMSs. It will contribute to a dialogue on investigating and prioritising learners’ ODL experiences to ensure education equity across all levels or categories of students, which aligns with the United Nations Sustainable Development Goals.Practical implicationsThe integration of UX and user testing allows us to better identify what users like, their concerns and their needs. We gain important input on how easy or difficult it is to use the system, move around it and how much they enjoy using it. This feedback helps us make changes to the design so that the final product is more in line with what users want. It also allows us to discover problems before they become major, saving time and effort later on. Finally, integrating user input improves the LMS, delivering a more fun and successful learning experience for everyone.Social implicationsUser-friendly systems arise as institutions prioritise user-centred design, breaking down barriers for various learners. This develops an innovative culture, improving present learning experiences and setting a precedent for future generations. The emphasis on user demands helps to create a more accessible, adaptive and egalitarian educational landscape by connecting education with current technological trends. As education becomes more inclusive, the broader community benefits, emphasising the beneficial social impact of LMS user testing.Originality/valueBy articulating the process of integrating user testing on an LMS/e-learning prototype, helps us understand what users like, where they face problems and what needs improvement. By involving users in testing, we get valuable feedback on how easy it is to use the system, navigate around and overall, how much they enjoy using it. Case studies like this also offer universities concrete examples of real-world challenges and successes.

Revisiting the current phyto-scape of Boomplaas Cave (South Africa) and the possible implications of this for past day-range foraging

ABSTRACT Boomplaas Cave in the Western Cape Province of South Africa is one of only a few African sites with inland archaeological deposits spanning Marine Isotope Stages 4–1. Work conducted half a century ago predicted Boomplaas to be a meagre plant-food location. We reassess this interpretation here by presenting updated lists of the current vegetation and foodplants growing within roughly a day’s foraging distance from the cave. By doing so, we increase the known foodplant species potentially available to Stone Age foragers by 356% and show that almost all the plant species/genera in the Boomplaas archaeobotanical assemblage still grow within a day’s range of the site. We present nutritional values for some of the plant foods, highlighting those richest in moisture, ash, protein, fat, fibre, carbohydrates and energy and suggesting that such foods may have been important staples in the dietary ecology of the Stone Age foragers who used the site. Lastly, we demonstrate that the Boomplaas Cave foodplant fitness landscape is relatively rich and varied compared to similar data from other Cape sites such as Klasies River Main Cave, Diepkloof Rock Shelter and Hollow Rock Shelter.

Distributed shape detection for an acoustic sensitive optical cable with DAS.

Acoustic sensitive optical cables (ASOCs) and their shape detection are vital in underwater acoustic monitoring, and a distributed ASOC shape detection method is demonstrated with distributed acoustic sensing (DAS) technology. The accurate three-dimensional (3D) position of each ASOC unit is obtained from DAS signals and the prior position information of auxiliary acoustic sources by using a proposed adaptive peak allocation algorithm. Preliminary work has demonstrated single-point 3D localization and distributed ASOC shape detection, and the error is 6.53 cm. To the best of our knowledge, it is the first time that distributed ASOC shape detection is achieved with DAS. This method will promote the development of ASOC applications, such as underwater target detection and towed array correction.

Evolutionary druggability for low-dimensional fitness landscapes toward new metrics for antimicrobial applications

The term ‘druggability’ describes the molecular properties of drugs or targets in pharmacological interventions and is commonly used in work involving drug development for clinical applications. There are no current analogues for this notion that quantify the drug-target interaction with respect to a given target variant’s sensitivity across a breadth of drugs in a panel, or a given drug’s range of effectiveness across alleles of a target protein. Using data from low-dimensional empirical fitness landscapes composed of 16 β-lactamase alleles and 7 β-lactam drugs, we introduce two metrics that capture (i) the average susceptibility of an allelic variant of a drug target to any available drug in a given panel (‘variant vulnerability’), and (ii) the average applicability of a drug (or mixture) across allelic variants of a drug target (‘drug applicability’). Finally, we (iii) disentangle the quality and magnitude of interactions between loci in the drug target and the seven drug environments in terms of their mutation by mutation by environment (G x G x E) interactions, offering mechanistic insight into the variant variability and drug applicability metrics. Summarizing, we propose that our framework can be applied to other datasets and pathogen-drug systems to understand which pathogen variants in a clinical setting are the most concerning (low variant vulnerability), and which drugs in a panel are most likely to be effective in an infection defined by standing genetic variation in the pathogen drug target (high drug applicability).

Evolutionary druggability for low-dimensional fitness landscapes toward new metrics for antimicrobial applications.

The term 'druggability' describes the molecular properties of drugs or targets in pharmacological interventions and is commonly used in work involving drug development for clinical applications. There are no current analogues for this notion that quantify the drug-target interaction with respect to a given target variant's sensitivity across a breadth of drugs in a panel, or a given drug's range of effectiveness across alleles of a target protein. Using data from low-dimensional empirical fitness landscapes composed of 16 β-lactamase alleles and 7 β-lactam drugs, we introduce two metrics that capture (i) the average susceptibility of an allelic variant of a drug target to any available drug in a given panel ('variant vulnerability'), and (ii) the average applicability of a drug (or mixture) across allelic variants of a drug target ('drug applicability'). Finally, we (iii) disentangle the quality and magnitude of interactions between loci in the drug target and the seven drug environments in terms of their mutation by mutation by environment (G x G x E) interactions, offering mechanistic insight into the variant variability and drug applicability metrics. Summarizing, we propose that our framework can be applied to other datasets and pathogen-drug systems to understand which pathogen variants in a clinical setting are the most concerning (low variant vulnerability), and which drugs in a panel are most likely to be effective in an infection defined by standing genetic variation in the pathogen drug target (high drug applicability).

Fitness Landscape Analysis of Product Unit Neural Networks

A fitness landscape analysis of the loss surfaces produced by product unit neural networks is performed in order to gain a better understanding of the impact of product units on the characteristics of the loss surfaces. The loss surface characteristics of product unit neural networks are then compared to the characteristics of loss surfaces produced by neural networks that make use of summation units. The failure of certain optimization algorithms in training product neural networks is explained through trends observed between loss surface characteristics and optimization algorithm performance. The paper shows that the loss surfaces of product unit neural networks have extremely large gradients with many deep ravines and valleys, which explains why gradient-based optimization algorithms fail at training these neural networks.