Bayesian Inference Approach Research Articles

Phylogenetic relationships of Neogene hamsters (Mammalia, Rodentia, Cricetinae) revealed under Bayesian inference and maximum parsimony.

There is an ongoing debate about the internal systematics of today's group of hamsters (Cricetinae), following new insights that are gained based on molecular data. Regarding the closely related fossil cricetids, however, most studies deal with only a limited number of genera and statements about their possible relationships are rare. In this study, 41 fossil species from the Late Miocene to the Pliocene, belonging to seven extinct cricetine genera, Collimys, Rotundomys, Neocricetodon, Pseudocricetus, Cricetulodon, Apocricetus and Hattomys are analysed in a phylogenetic framework using traditional maximum parsimony and Bayesian inference approaches. Following thorough model testing, a relaxed-clock Bayesian inference analysis is performed under tip-dating to estimate divergence times simultaneously. Furthermore, so-called 'rogue' taxa are identified and excluded from the final trees to improve the informative value of the shown relationships. Based on these resulting trees, the fit of the topologies to the stratigraphy is assessed and the ancestral states of the characters are reconstructed under a parsimonious approach and stochastic character mapping. The overall topologies resulting from Bayesian and parsimonious approaches are largely congruent to each other and confirm the monophyly of most of the genera. Additionally, synapomorphies can be identified for each of these genera based on the ancestral state reconstructions. Only Cricetulodon turns out to be paraphyletic, while 'Cricetulodon' complicidens is a member of Neocricetodon. Lastly, this work makes a contribution to a debate that went on for decades, as the genus Kowalskia can be confirmed as junior synonym of Neocricetodon.

Comparative analysis of complete chloroplast genomes of 14 Asteraceae species.

The Asteraceae family, the largest and one of the most diverse families of angiosperms, presents significant challenges in taxonomic classification and systematic research due to its vast species diversity and morphological complexity. A comprehensive understanding of the chloroplast genomes within this family is essential for refining taxonomic classifications and advancing phylogenetic studies. In this study, we sequenced the complete chloroplast genomes of 14 Asteraceae species and conducted a thorough bioinformatic analysis of their characteristics. The chloroplast genomes, ranging from 150,907bp to 152,858bp, exhibit a typical quadripartite structure: a large single-copy (LSC) region (83,044bp to 84,625bp), a small single-copy (SSC) region (18,223bp to 18,673bp), and a pair of inverted repeats (IRs) (24,806bp to 25,201bp). These genomes encode 87 to 89 protein-coding genes (PCGs), 36 to 37 tRNA genes, and 8 rRNA genes, with high conservation in size, structure, gene content, and order. Comparative analysis with other Asteraceae species' chloroplast genomes revealed notable similarities and structural variations, particularly in the IR regions. Nucleotide polymorphism analysis identified four genes-trnY-GUA, trnE-UUC, ycf1, and rrn23-with higher Pi values, suggesting potential hotspots for evolutionary studies. Phylogenetic analysis using maximum likelihood (ML) and Bayesian inference (BI) approaches provided new insights, proposing the reclassification of Himalaiella auriculata and Jacobaea raphanifolia as independent genera, distinct from Saussurea and Senecio. This study presents a comprehensive analysis of the chloroplast genome structures and phylogenetic relationships of 14 Asteraceae species, offering critical data for future molecular identification, evolutionary biology, and population genetics research. The findings hold significant implications for the ongoing refinement of Asteraceae taxonomic classifications and enhance our understanding of the evolutionary dynamics within this complex family.

Bayesian blockwise inference for joint models of longitudinal and multistate data with application to longitudinal multimorbidity analysis.

Multistate models provide a useful framework for modelling complex event history data in clinical settings and have recently been extended to the joint modelling framework to appropriately handle endogenous longitudinal covariates, such as repeatedly measured biomarkers, which are informative about health status and disease progression. However, the practical application of such joint models faces considerable computational challenges. Motivated by a longitudinal multimorbidity analysis of large-scale UK health records, we introduce novel Bayesian inference approaches for these models that are capable of handling complex multistate processes and large datasets with straightforward implementation. These approaches decompose the original estimation task into smaller inference blocks, leveraging parallel computing and facilitating flexible model specification and comparison. Using extensive simulation studies, we show that the proposed approaches achieve satisfactory estimation accuracy, with notable gains in computational efficiency compared to the standard Bayesian estimation strategy. We illustrate our approaches by analysing the coevolution of routinely measured systolic blood pressure and the progression of three important chronic conditions, using a large dataset from the Clinical Practice Research Datalink Aurum database. Our analysis reveals distinct and previously lesser-known association structures between systolic blood pressure and different disease transitions.

Bayesian inference of state feedback control parameters for fo perturbation responses in cerebellar ataxia.

Behavioral speech tasks have been widely used to understand the mechanisms of speech motor control in typical speakers as well as in various clinical populations. However, determining which neural functions differ between typical speakers and clinical populations based on behavioral data alone is difficult because multiple mechanisms may lead to the same behavioral differences. For example, individuals with cerebellar ataxia (CA) produce atypically large compensatory responses to pitch perturbations in their auditory feedback, compared to typical speakers, but this pattern could have many explanations. Here, computational modeling techniques were used to address this challenge. Bayesian inference was used to fit a state feedback control (SFC) model of voice fundamental frequency (fo) control to the behavioral pitch perturbation responses of speakers with CA and typical speakers. This fitting process resulted in estimates of posterior likelihood distributions for five model parameters (sensory feedback delays, absolute and relative levels of auditory and somatosensory feedback noise, and controller gain), which were compared between the two groups. Results suggest that the speakers with CA may proportionally weight auditory and somatosensory feedback differently from typical speakers. Specifically, the CA group showed a greater relative sensitivity to auditory feedback than the control group. There were also large group differences in the controller gain parameter, suggesting increased motor output responses to target errors in the CA group. These modeling results generate hypotheses about how CA may affect the speech motor system, which could help guide future empirical investigations in CA. This study also demonstrates the overall proof-of-principle of using this Bayesian inference approach to understand behavioral speech data in terms of interpretable parameters of speech motor control models.

Mitochondrial data unravels the human-mediated introduction of Mediodactylus kotschyi (Squamata: Gekkonidae) to the Central Balkan Peninsula

Abstract Kotschy’s gecko, Mediodactylus kotschyi, is native to the Eastern part of the Mediterranean Basin and a successful colonizer, with an extensive distribution across the Balkans and outside its native range. In Serbia, it was presumed native in Prizren (Kosovo and Metohija province) and introduced across the remaining territory. To explore the origin of Serbian Kotschy’s gecko, we collected specimens from most of the Serbian populations where the species was recorded and assessed their phylogenetic relationships based on three mtDNA markers (COI, cytb, 16S rRNA) under Maximum Likelihood (ML) and Bayesian Inference (BI) approaches. First, our results support that the specimens found in Serbia belong to M. kotschyi. Second, all populations except Prizren belong to the subspecies M. kotschyi skopjensis. Furthermore, the Italian populations are sister to the specimens from Prizren, suggesting a Balkan origin for these populations, which settled in Apulia likely through ancient human-mediated dispersion. The small genetic differentiation found suggests that Kotschy’s gecko rapidly colonized the Central Balkans, originating from Northern Greece and North Macedonia. The results also suggest a story of multiple, complex (most likely) human-mediated introductions. However, further studies should be conducted to properly elucidate the colonization pathways, and to truly disentangle the potential fast natural range expansion from an anthropogenic dispersion of this gecko species.

Bayesian Probabilistic Inference of Nonparametric Distance Distributions in DEER Spectroscopy.

Double electron-electron resonance (DEER) spectroscopy measures distance distributions between spin labels in proteins, yielding important structural and energetic information about conformational landscapes. Analysis of an experimental DEER signal in terms of a distance distribution is a nontrivial task due to the ill-posed nature of the underlying mathematical inversion problem. This work introduces a Bayesian probabilistic inference approach to analyze DEER data, assuming a nonparametric distance distribution with a Tikhonov smoothness prior. The method uses Markov Chain Monte Carlo sampling with a compositional Gibbs sampler to determine a posterior probability distribution over the entire parameter space, including the distance distribution, given an experimental data set. This posterior contains all of the information available from the data, including a full quantification of the uncertainty about the model parameters. The corresponding uncertainty about the distance distribution is visually captured via an ensemble of posterior predictive distributions. Several examples are presented to illustrate the method. Compared with bootstrapping, it performs faster and provides slightly larger uncertainty intervals.

Advancing human mobility modeling: a novel path flow approach to mining traffic congestion dynamics

Mining traffic congestion dynamics presents difficulties in data structure and spatiotemporal analysis. Existing studies mainly provide insights from a supply perspective, with a restricted examination of why congestion occurs and how travel demands affect congestion. This study introduces an innovative framework to mine congestion dynamics from the perspective of human mobility. In human mobility modeling, we refine the conventional origin-destination representation of activity flow by introducing "path flow" (PF) which considers space-time paths and movement patterns. In congestion scenarios, congestion-related path flow (CPF) and congested path sub-flow (CPSF) are extended to track individuals’ congestion exposure and explore the correlations between congestion and human mobility. To finely classify congestion-related travel demands, a Bayesian inference approach, incorporating destination and spatiotemporal heterogeneities, is developed to deduce trip purposes. The experiments conducted in Wuhan demonstrate the availability and importance of PF in spatiotemporal dynamics analysis of human mobility. Interestingly, we find that 1) the job-housing relationship is imbalanced, with massive residents opting for cross-district living and working; 2) individuals tend to visit tertiary hospitals on weekends and secondary medical facilities with less congestion on weekdays. Notably, path flow can promote the fine-grained modeling of human mobility and provide theoretical support for many urban issues.

Brain sodium MRI-derived priors support the estimation of epileptogenic zones using personalized model-based methods in epilepsy.

Patients presenting with drug-resistant epilepsy are eligible for surgery aiming to remove the regions involved in the production of seizure activities, the so-called epileptogenic zone network (EZN). Thus the accurate estimation of the EZN is crucial. Data-driven, personalized virtual brain models derived from patient-specific anatomical and functional data are used in Virtual Epileptic Patient (VEP) to estimate the EZN via optimization methods from Bayesian inference. The Bayesian inference approach used in previous VEP integrates priors, based on the features of stereotactic-electroencephalography (SEEG) seizures' recordings. Here, we propose new priors, based on quantitative 23Na-MRI. The 23Na-MRI data were acquired at 7T and provided several features characterizing the sodium signal decay. The hypothesis is that the sodium features are biomarkers of neuronal excitability related to the EZN and will add additional information to VEP estimation. In this paper, we first proposed the mapping from 23Na-MRI features to predict the EZN via a machine learning approach. Then, we exploited these predictions as priors in the VEP pipeline. The statistical results demonstrated that compared with the results from current VEP, the result from VEP based on 23Na-MRI prior has better balanced accuracy, and the similar weighted harmonic mean of the precision and recall.

A Bayesian Network Inference Approach for Dynamic Risk Assessment Using Multisource-Based Information Fusion in an Interval Type-2 Fuzzy Set Environment

A Bayesian Network Inference Approach for Dynamic Risk Assessment Using Multisource-Based Information Fusion in an Interval Type-2 Fuzzy Set Environment

Nonparametric Bayesian reconstruction of Galactic magnetic fields using information field theory

Context. Ultrahigh-energy cosmic rays (UHECRs) are charged particles with energies surpassing 1018 eV. Their sources remain elusive because they are obscured by deflections caused by the Galactic magnetic field (GMF). This challenge is further complicated by our limited understanding of the 3D structure of the GMF because current GMF observations primarily consist of quantities that are integrated along the line of sight (LOS). Nevertheless, data from upcoming stellar polarization surveys along with Gaia stellar parallax data are expected to yield local GMF measurements. Aims. This study is the second entry in our exploration of a Bayesian inference approach to the local GMF that uses synthetic local GMF observations that emulate forthcoming local GMF measurements, and attempts to use them to reconstruct its 3D structure. The ultimate aim is to trace back observed UHECRs and thereby update our knowledge about their possible origin. Methods. In this proof-of-concept work, we assumed as ground truth a magnetic field produced by a dynamo simulation of the Galactic ISM. We employed methods of Bayesian statistical inference in order to sample the posterior distribution of the GMF within part of the Galaxy. By assuming a known rigidity and arrival direction of an UHECR, we traced its trajectory back through various GMF configurations drawn from the posterior distribution. Our objective was to rigorously evaluate the performance of our algorithm in scenarios that closely mirror the setting of expected future applications. In pursuit of this, we conditioned the posterior to synthetically integrated LOS measurements of the GMF, in addition to synthetic local plane of sky-component measurements. Results. Our results demonstrate that for all locations of the observed arrival direction on the plane of sky, our algorithm is able to substantially update our knowledge on the original arrival direction of UHECRs with a rigidity of E/Z = 5 × 1019 eV, even without any LOS information. When the integrated data are included in the inference, the regions of the celestial sphere in which the maximum error occurs are greatly reduced. The maximum error is diminished by a factor of about 3 even in these regions in the specific setting we studied. Additionally, we are able to identify the regions in which the largest error is expected to occur.

A vaccination-based COVID-19 model: Analysis and prediction using Hamiltonian Monte Carlo

Compartmental models have emerged as robust computational frameworks and have yielded remarkable success in the fight against COVID-19. This study proposes a vaccination-based compartmental model for COVID-19 transmission dynamics. The model reflects the specific stages of COVID-19 infection and integrates a vaccination strategy, allowing for a comprehensive analysis of how vaccination rates influence the disease spread. We fit this model to daily confirmed COVID-19 cases in Tennessee, United States of America (USA), from June 4 to November 26, 2021, in a Bayesian inference approach using the Hamiltonian Monte Carlo (HMC) algorithm. First, excluding vaccination dynamics from the model, we estimated key epidemiological parameters like infection, recovery, and disease-induced death rates. This analysis yielded a basic reproduction number (R0) of 1.5. Second, we incorporated vaccination dynamics and estimated the vaccination rate for three vaccines: 0.0051 per day for both Pfizer and Moderna and 0.0059 per day for Janssen. The fitted curves show reductions in the epidemic peak for all three vaccines. Pfizer and Moderna vaccines bring the peak down from 8,029 infected cases to 5,616 infected cases, while the Janssen vaccine reduces it, to 6,493 infected cases. Simulations of the model by varying the vaccination rate and vaccine efficacy were performed. A highly effective vaccine (95% efficacy) with a daily vaccination rate of 0.006 halved COVID-19 infections, reducing cases from 8,029 to around 4,000. The results also show that the model's prediction accuracy for new observations improves with the number of observed data used to train the model.

Synthetic Likelihood in Misspecified Models

Bayesian synthetic likelihood is a widely used approach for conducting Bayesian analysis in complex models where evaluation of the likelihood is infeasible but simulation from the assumed model is tractable. We analyze the behavior of the Bayesian synthetic likelihood posterior when the assumed model differs from the actual data generating process. We demonstrate that the Bayesian synthetic likelihood posterior can display a wide range of nonstandard behaviors depending on the level of model misspecification, including multimodality and asymptotic non-Gaussianity. Our results suggest that likelihood tempering, a common approach for robust Bayesian inference, fails for synthetic likelihood whilst recently proposed robust synthetic likelihood approaches can ameliorate this behavior and deliver reliable posterior inference under model misspecification. All results are illustrated using a simple running example. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Epidemiological modeling of COVID-19 data with Advanced statistical inference based on Type-II progressive censoring

This research proposes the Kavya-Manoharan Unit Exponentiated Half Logistic (KM-UEHL) distribution as a novel tool for epidemiological modeling of COVID-19 data. Specifically designed to analyze data constrained to the unit interval, the KM-UEHL distribution builds upon the unit exponentiated half logistic model, making it suitable for various data from COVID-19. The paper emphasizes the KM-UEHL distribution's adaptability by examining its density and hazard rate functions. Its effectiveness is demonstrated in handling the diverse nature of COVID-19 data through these functions. Key characteristics like moments, quantile functions, stress-strength reliability, and entropy measures are also comprehensively investigated. Furthermore, the KM-UEHL distribution is employed for forecasting future COVID-19 data under a progressive Type-II censoring scheme, which acknowledges the time-dependent nature of data collection during outbreaks. The paper presents various methods for constructing prediction intervals for future-order statistics, including maximum likelihood estimation, Bayesian inference (both point and interval estimates), and upper-order statistics approaches. The Metropolis-Hastings and Gibbs sampling procedures are combined to create the Markov chain Monte Carlo simulations because it is mathematically difficult to acquire closed-form solutions for the posterior density function in the Bayesian framework. The theoretical developments are validated with numerical simulations, and the practical applicability of the KM-UEHL distribution is showcased using real-world COVID-19 datasets.

Characteristics and Comparative Analysis of Six Mitogenomes of Genus Kiefferulus Goetghebuer, 1922 (Diptera: Chironomidae).

Chironomidae is a cosmopolitan and species-rich family of insects, with many species serving as useful indicators of aquatic ecosystem health. In this study, we newly sequenced six species of Kiefferulus Goetghebuer, 1922 (Chironomidae: Chironominae) by high-throughput sequencing technology. We analyzed characters of the mitochondrial genome, including the sequence length, nucleotide composition, and evolutionary rates of this genus. The size of the newly obtained sequences ranged from 15,588 to 15,767 bp, and all of them included 22 tRNAs, 13 PCGs, 2 rRNAs, and 1 CR. The CR showed the highest AT content relative to the PCGs, rRNAs, and tRNAs. Relative synonymous codon usage analysis showed that UUA, UUU, and AUU are the preferred codons. The ratio of nonsynonymous (Ka) to synonymous (Ks) substitution rates showed that all Ka/Ks of PCGs were lower than 1, with ATP8 having the highest evolution rate, while COX1 exhibited the lowest evolution rate. We reconstructed the phylogenetic relationship of the genus Kiefferulus based on eight species (six ingroups and two outgroups), using five matrices and employing Maximum likelihood and Bayesian inference approaches. Phylogenetic analysis of the Kiefferulus showed that six species within this genus were classified into a monophyletic clade.

Neural Operator Variational Inference Based on Regularized Stein Discrepancy for Deep Gaussian Processes.

Deep Gaussian process (DGP) models offer a powerful nonparametric approach for Bayesian inference, but exact inference is typically intractable, motivating the use of various approximations. However, existing approaches, such as mean-field Gaussian assumptions, limit the expressiveness and efficacy of DGP models, while stochastic approximation can be computationally expensive. To tackle these challenges, we introduce neural operator variational inference (NOVI) for DGPs. NOVI uses a neural generator to obtain a sampler and minimizes the regularized Stein discrepancy (RSD) between the generated distribution and true posterior in L2 space. We solve the minimax problem using Monte Carlo estimation and subsampling stochastic optimization techniques and demonstrate that the bias introduced by our method can be controlled by multiplying the Fisher divergence with a constant, which leads to robust error control and ensures the stability and precision of the algorithm. Our experiments on datasets ranging from hundreds to millions demonstrate the effectiveness and the faster convergence rate of the proposed method. We achieve a classification accuracy of 93.56 on the CIFAR10 dataset, outperforming state-of-the-art (SOTA) Gaussian process (GP) methods. We are optimistic that NOVI possesses the potential to enhance the performance of deep Bayesian nonparametric models and could have significant implications for various practical applications.

Molecular Phylogenetics and Phylogeography of Malaysian Mousedeer (Tragulus kanchil) Based on Mitochondrial DNA Sequences of the D-Loop Region

Two sympatric mousedeer species, Tragulus kanchil and Tragulus napu, inhabit the Malaysian tropical rainforests. Previous studies have established their phylogenetic relationships based on morphological variations; however, comprehensive genetic data have yet to be extensively used to relate the relationships especially from different populations. The main objectives of this study were to determine the phylogenetic relationships, population genetics, and phylogeography of mousedeer species based on DNA sequences of the mitochondrial D-loop region. DNA from 32 mousedeer samples, representing various populations in Malaysia, was sequenced and analyzed using Neighbor-Joining, Maximum Parsimony, and Bayesian Inference approaches. The phylogenetic analyses showed two main clades representing the populations of T. kanchil and T. napu. The results also showed that the T. kanchil populations in Borneo was separated from Peninsular Malaysia taxa in MP and BI phylogenetic tree. However, Borneo population was nested in east population of Peninsular Malaysia in NJ tree. In addition, the T. kanchil population in Peninsular Malaysia further separated into east and west coast populations of Titiwangsa Range. It was supported with finding in population genetic relation that showed relatively low levels among population. This is expected because some of the populations are isolated geographically. The divergence of these populations is likely due to the Titiwangsa Range which acts as a barrier separating the east and west Peninsular Malaysia populations, and the South China Sea separates the populations of Peninsular Malaysia and Borneo. Molecular clock tree reconstruction showed that the separation of T. kanchil and T. napu occurred around 17.11 million years ago (MYAs). Furthermore, the T. kanchil populations from the east and west Peninsular Malaysia showed a branching pattern from those of Borneo of about 11.04 and 9.14 MYAs, respectively. The results of this study increase our understanding of Malaysian mousedeer phylogeny and phylogeography.

Software reliability prediction: A machine learning and approximation Bayesian inference approach

AbstractReliability growth models are commonly categorized into two primary groups: parametric and non‐parametric models. Parametric models, known as Software Reliability Growth Models (SRGM) rely on a set of hypotheses that can potentially affect the accuracy of model predictions, while non‐parametric models (such as neural networks) can predict the model solely based on training data without any assumptions regarding the model itself. In this paper, we propose several methods to enhance prediction accuracy in software reliability context. More specifically, we, on one hand, introduce two gradient‐based techniques for estimating parameters of classical SRGMs. On the other, we propose methods involving LSTM Encoder–Decoder and Bayesian approximation within Langevin Gradient and Variational inference neural networks. To evaluate our proposed models' performance, we compare them with various neural network‐based software reliability models using three real‐world software failure datasets and utilizing the Mean Square Error (MSE) as a model comparison criterion. The experimental results indicate that our proposed non‐parametric models outperform most classical parametric and non‐parametric models.

A trip of a bamboo lifetime: tracing the origin and evolution of Chusquea (Poaceae, Bambusoideae, Bambuseae) in Patagonia, southernmost South America

Abstract We performed a phylogenetic analysis with estimated clade ages and an ancestral-range estimation focused on the Patagonian species of Chusquea, i.e. the Chusquea culeou clade, to unravel their biogeographical history. Phylogenetic analyses were conducted with sequence data from four plastid markers, using maximum parsimony and Bayesian inference approaches. The complete sampling and addition of the Patagonian Chusquea species into a phylogenetic framework revealed the most likely origin and evolutionary history of the Chusquea culeou clade. The Patagonian Chusquea originated roughly 11.35 Mya, most likely from a common ancestor from the Atlantic Forest region, when this lineage underwent in situ diversification and its species evolved sympatrically. All the extant species have arisen in the last 8 Myr. Moreover, according to the present sampling, the most recent common ancestor of subtribe Chusqueinae originated in the Northern Andes, which is also the most common source area of the estimated dispersal events. Some other preliminary insights on the biogeographic history of the genus are included.

Accurate model and ensemble refinement using cryo-electron microscopy maps and Bayesian inference.

Converting cryo-electron microscopy (cryo-EM) data into high-quality structural models is a challenging problem of outstanding importance. Current refinement methods often generate unbalanced models in which physico-chemical quality is sacrificed for excellent fit to the data. Furthermore, these techniques struggle to represent the conformational heterogeneity averaged out in low-resolution regions of density maps. Here we introduce EMMIVox, a Bayesian inference approach to determine single-structure models as well as structural ensembles from cryo-EM maps. EMMIVox automatically balances experimental information with accurate physico-chemical models of the system and the surrounding environment, including waters, lipids, and ions. Explicit treatment of data correlation and noise as well as inference of accurate B-factors enable determination of structural models and ensembles with both excellent fit to the data and high stereochemical quality, thus outperforming state-of-the-art refinement techniques. EMMIVox represents a flexible approach to determine high-quality structural models that will contribute to advancing our understanding of the molecular mechanisms underlying biological functions.

Large Skew-t Copula Models and Asymmetric Dependence in Intraday Equity Returns

Skew-t copula models are attractive for the modeling of financial data because they allow for asymmetric and extreme tail dependence. We show that the copula implicit in the skew-t distribution of Azzalini and Capitanio allows for a higher level of pairwise asymmetric dependence than two popular alternative skew-t copulas. Estimation of this copula in high dimensions is challenging, and we propose a fast and accurate Bayesian variational inference (VI) approach to do so. The method uses a generative representation of the skew-t distribution to define an augmented posterior that can be approximated accurately. A stochastic gradient ascent algorithm is used to solve the variational optimization. The methodology is used to estimate skew-t factor copula models with up to 15 factors for intraday returns from 2017 to 2021 on 93 U.S. equities. The copula captures substantial heterogeneity in asymmetric dependence over equity pairs, in addition to the variability in pairwise correlations. In a moving window study we show that the asymmetric dependencies also vary over time, and that intraday predictive densities from the skew-t copula are more accurate than those from benchmark copula models. Portfolio selection strategies based on the estimated pairwise asymmetric dependencies improve performance relative to the index.