Robust SNP-based prediction of rheumatoid arthritis through machine-learning-optimized polygenic risk score.

BACKGROUND: The popular statistics-based Genome-wide association studies (GWAS) have provided deep insights into the field of complex disorder genetics. However, its clinical applicability to predict disease/trait outcomes remains unclear as statistical models are not designed to make predictions. This study employs statistics-free machine-learning (ML)-optimized polygenic risk score (PRS) to complement existing GWAS and bring the prediction of disease/trait outcomes closer to clinical application. Rheumatoid Arthritis (RA) was selected as a model disease to demonstrate the robustness of ML in disease prediction as RA is a prevalent chronic inflammatory joint disease with high mortality rates, affecting adults at the economic prime. Early identification of at-risk individuals may facilitate measures to mitigate the effects of the disease. METHODS: This study employs a robust ML feature selection algorithm to identify single nucleotide polymorphisms (SNPs) that can predict RA from a set of training data comprising RA patients and population control samples. Thereafter, selected SNPs were evaluated for their predictive performances across 3 independent, unseen test datasets. The selected SNPs were subsequently used to generate PRS which was also evaluated for its predictive capacity as a sole feature. RESULTS: Through robust ML feature selection, 9 SNPs were found to be the minimum number of features for excellent predictive performance (AUC > 0.9) in 3 independent, unseen test datasets. PRS based on these 9 SNPs was significantly associated with (P < 1 × 10-16) and predictive (AUC > 0.9) of RA in the 3 unseen datasets. A RA ML-PRS calculator of these 9 SNPs was developed ( https://xistance.shinyapps.io/prs-ra/ ) to facilitate individualized clinical applicability. The majority of the predictive SNPs are protective, reside in non-coding regions, and are either predicted to be potentially functional SNPs (pfSNPs) or in high linkage disequilibrium (r2 > 0.8) with un-interrogated pfSNPs. CONCLUSIONS: These findings highlight the promise of this ML strategy to identify useful genetic features that can robustly predict disease and amenable to translation for clinical application.

Self-organization and information transfer in Antarctic krill swarms.

Antarctic krill swarms are one of the largest known animal aggregations, and yet, despite being the keystone species of the Southern Ocean, little is known about how swarms are formed and maintained. Understanding the local interactions between individuals that provide the basis for these swarms is fundamental to knowing how swarms arise in nature, and what potential factors might lead to their breakdown. Here, we analysed the trajectories of captive, wild-caught krill in 3D to determine individual-level interaction rules and quantify patterns of information flow. Our results demonstrate that krill align with near neighbours and that they regulate both their direction and speed relative to the positions of groupmates. These results suggest that social factors are vital to the formation and maintenance of swarms. Furthermore, krill operate a novel form of collective organization, with measures of information flow and individual movement adjustments expressed most strongly in the vertical dimension, a finding not seen in other swarming species. This research represents a vital step in understanding the fundamentally important swarming behaviour of krill.

Machine learning using genetic and clinical data identifies a signature that robustly predicts methotrexate response in rheumatoid arthritis.

OBJECTIVE: To develop a hypothesis-free model that best predicts response to MTX drug in RA patients utilizing biologically meaningful genetic feature selection of potentially functional single nucleotide polymorphisms (pfSNPs) through robust machine learning (ML) feature selection methods. METHODS: MTX-treated RA patients with known response were divided in a 4:1 ratio into training and test sets. From the patients' exomes, potential features for classifier prediction were identified from pfSNPs and non-genetic factors through ML using recursive feature elimination with cross-validation incorporating the random forest classifier. Feature selection was repeated on random subsets of the training cohort, and consensus features were assembled into the final feature set. This feature set was evaluated for predictive potential using six ML classifiers, first by cross-validation within the training set, and finally by analysing its performance with the unseen test set. RESULTS: The final feature set contains 56 pfSNPs and five non-genetic factors. The majority of these pfSNPs are located in pathways related to RA pathogenesis or MTX action and are predicted to modulate gene expression. When used for training in six ML classifiers, performance was good in both the training set (area under the curve: 0.855-0.916; sensitivity: 0.715-0.892; and specificity: 0.733-0.862) and the unseen test set (area under the curve: 0.751-0.826; sensitivity: 0.581-0.839; and specificity: 0.641-0.923). CONCLUSION: Sensitive and specific predictors of MTX response in RA patients were identified in this study through a novel strategy combining biologically meaningful and machine learning feature selection and training. These predictors may facilitate better treatment decision-making in RA management.

Fine-scale behavioural adjustments of prey on a continuum of risk.

In the wild, prey species often live in the vicinity of predators, rendering the ability to assess risk on a moment-to-moment basis crucial to survival. Visual cues are important as they allow prey to assess predator species, size, proximity and behaviour. However, few studies have explicitly examined prey's ability to assess risk based on predator behaviour and orientation. Using mosquitofish, Gambusia holbrooki, and their predator, jade perch, Scortum barcoo, under controlled conditions, we provide some of the first fine-scale characterization of how prey adapt their behaviour according to their continuous assessment of risk based on both predator behaviour and angular distance to the predator's mouth. When these predators were inactive and posed less of an immediate threat, prey within the attack cone of the predator showed reductions in speed and acceleration characteristic of predator-inspection behaviour. However, when predators became active, prey swam faster with greater acceleration and were closer together within the attack cone of predators. Most importantly, this study provides evidence that prey do not adopt a uniform response to the presence of a predator. Instead, we demonstrate that prey are capable of rapidly and dynamically updating their assessment of risk and showing fine-scale adjustments to their behaviour.

Mid-sized groups perform best in a collective decision task in sticklebacks.

Numerous studies have reported functional improvements in collective behaviour with increasing group size, however, the possibility that such improvements may saturate or even decline as group size continues to grow have seldom been tested experimentally. Here, we tested the ability of solitary three-spined sticklebacks and those in groups, ranging from 2 to 29 fish, to leave an unfavourable patch of habitat. Our results replicate the findings of previous studies at low group sizes, with the fish initially showing a reduction in their latency to leave the unfavourable habitat as group size increased. As group size continued to increase, however, latency to leave the habitat increased, so that the functional relationship between group size and latency to depart was U-shaped. Our results suggest an optimum group size in this context of between 12 and 20 fish. Underlying this group-level trend was a similar U-shaped relationship between group size and the first fish to leave the habitat, suggesting that at larger group sizes, social conformity to the behaviour of the majority can stifle the ability of fish to innovate-in this case, to induce a collective movement from the unfavourable habitat.

Personality affects the foraging response of a mammalian herbivore to the dual costs of food and fear.

Predators attack and plants defend, so herbivores face the dilemma of how to eat enough without being eaten. But do differences in the personality of herbivores affect the foraging choices of individuals? We explored the ecological impact of personality in a generalist herbivore, the brushtail possum (Trichosurus vulpecula). After quantifying personality traits in wild individuals brought temporarily into captivity, we tested how these traits altered foraging by individuals when free-ranging in their natural habitat. To measure their responses to the dual costs of predation risk and plant toxin, we varied the toxin concentration of food in safe foraging patches against paired, non-toxic risky patches, and used a novel synthesis of a manipulative Giving-Up-Density (GUD) experiment and video behavioural analysis. At the population level, the cost of safe patches pivoted around that of risky patches depending on food toxin concentration. At the individual level, boldness affected foraging at risky high-quality food patches (as behavioural differences between bold and shy), and at safe patches only when food toxin concentration was low (as differences in foraging outcome). Our results ecologically validate the personality trait of boldness, in brushtail possums. They also reveal, for the first time, a nuanced link between personality and the way in which individuals balance the costs of food and fear. Importantly, they suggest that high plant defence effectively attenuates differences in foraging behaviour arising from variation in personality, but poorly defended plants in safe areas should be differentially subject to herbivory depending on the personality of the herbivore.

Multi-scale inference of interaction rules in animal groups using Bayesian model selection.

Inference of interaction rules of animals moving in groups usually relies on an analysis of large scale system behaviour. Models are tuned through repeated simulation until they match the observed behaviour. More recent work has used the fine scale motions of animals to validate and fit the rules of interaction of animals in groups. Here, we use a Bayesian methodology to compare a variety of models to the collective motion of glass prawns (Paratya australiensis). We show that these exhibit a stereotypical 'phase transition', whereby an increase in density leads to the onset of collective motion in one direction. We fit models to this data, which range from: a mean-field model where all prawns interact globally; to a spatial Markovian model where prawns are self-propelled particles influenced only by the current positions and directions of their neighbours; up to non-Markovian models where prawns have 'memory' of previous interactions, integrating their experiences over time when deciding to change behaviour. We show that the mean-field model fits the large scale behaviour of the system, but does not capture the observed locality of interactions. Traditional self-propelled particle models fail to capture the fine scale dynamics of the system. The most sophisticated model, the non-Markovian model, provides a good match to the data at both the fine scale and in terms of reproducing global dynamics, while maintaining a biologically plausible perceptual range. We conclude that prawns' movements are influenced by not just the current direction of nearby conspecifics, but also those encountered in the recent past. Given the simplicity of prawns as a study system our research suggests that self-propelled particle models of collective motion should, if they are to be realistic at multiple biological scales, include memory of previous interactions and other non-Markovian effects.

Fast and accurate decisions through collective vigilance in fish shoals.

Although it has been suggested that large animal groups should make better decisions than smaller groups, there are few empirical demonstrations of this phenomenon and still fewer explanations of the how these improvements may be made. Here we show that both speed and accuracy of decision making increase with group size in fish shoals under predation threat. We examined two plausible mechanisms for this improvement: first, that groups are guided by a small proportion of high-quality decision makers and, second, that group members use self-organized division of vigilance. Repeated testing of individuals showed no evidence of different decision-making abilities between individual fish. Instead, we suggest that shoals achieve greater decision-making efficiencies through division of labor combined with social information transfer. Our results should prompt reconsideration of how we view cooperation in animal groups with fluid membership.

Inferring the rules of interaction of shoaling fish.

Collective motion, where large numbers of individuals move synchronously together, is achieved when individuals adopt interaction rules that determine how they respond to their neighbors' movements and positions. These rules determine how group-living animals move, make decisions, and transmit information between individuals. Nonetheless, few studies have explicitly determined these interaction rules in moving groups, and very little is known about the interaction rules of fish. Here, we identify three key rules for the social interactions of mosquitofish (Gambusia holbrooki): (i) Attraction forces are important in maintaining group cohesion, while we find only weak evidence that fish align with their neighbor's orientation; (ii) repulsion is mediated principally by changes in speed; (iii) although the positions and directions of all shoal members are highly correlated, individuals only respond to their single nearest neighbor. The last two of these rules are different from the classical models of collective animal motion, raising new questions about how fish and other animals self-organize on the move.

Collective order and group structure of shoaling fish subject to differing risk-level treatments with a sympatric predator.

It is imperative for individuals to exhibit flexible behaviour according to ecological context, such as available resources or predation threat. Manipulative studies on responses to threat often focus on behaviour in the presence of a single indicator for the potential of predation, whereas in the wild perception of threat will probably be more nuanced. Here, we examine the collective behaviour of eastern mosquitofish (Gambusia holbrooki) subject to five differing threat scenarios relating to the presence and hunger state of a jade perch (Scortum barcoo). Across threat scenarios, groups exhibit unique behavioural profiles that differ in the durations that particular collective states are maintained, the probability of transitions between states, the size and duration of persistence of spatially defined subgroups, and the patterns of collective order of these subgroups. Under the greatest level of threat, subgroups of consistent membership persist for longer durations. Group-level behaviours, and their differences, are interconnected with differences in estimates of the underlying rules of interaction thought to govern collective motion. The responses of the group are shown to be specific to the details of a potential threat, rather than a binary response to the presence or absence of some form of threat.

Initiators, leaders, and recruitment mechanisms in the collective movements of damselfish.

Explaining how individual behavior and social interactions give rise to group-level outcomes and affect issues such as leadership is fundamental to the understanding of collective behavior. Here we examined individual and collective behavioral dynamics in groups of humbug damselfish both before and during a collective movement. During the predeparture phase, group activity increased until the collective movement occurred. Although such movements were precipitated by one individual, the success or failure of any attempt to instigate a collective movement was not solely dependent on this initiator's behavior but on the behavior of the group as a whole. Specifically, groups were more active and less cohesive before a successful initiation attempt than before a failed attempt. Individuals who made the most attempts to initiate a collective movement during each trial were ultimately most likely to lead the collective movement. Leadership was not related to dominance but was consistent between trials. The probability of fish recruiting to a group movement initiative was an approximately linear function of the number of fish already recruited. Overall, these results are consistent with nonselective local mimetism, with the decision to leave based on a group's, rather than any particular individual's, readiness to leave.

Accurate decisions in an uncertain world: collective cognition increases true positives while decreasing false positives.

In a wide range of contexts, including predator avoidance, medical decision-making and security screening, decision accuracy is fundamentally constrained by the trade-off between true and false positives. Increased true positives are possible only at the cost of increased false positives; conversely, decreased false positives are associated with decreased true positives. We use an integrated theoretical and experimental approach to show that a group of decision-makers can overcome this basic limitation. Using a mathematical model, we show that a simple quorum decision rule enables individuals in groups to simultaneously increase true positives and decrease false positives. The results from a predator-detection experiment that we performed with humans are in line with these predictions: (i) after observing the choices of the other group members, individuals both increase true positives and decrease false positives, (ii) this effect gets stronger as group size increases, (iii) individuals use a quorum threshold set between the average true- and false-positive rates of the other group members, and (iv) individuals adjust their quorum adaptively to the performance of the group. Our results have broad implications for our understanding of the ecology and evolution of group-living animals and lend themselves for applications in the human domain such as the design of improved screening methods in medical, forensic, security and business applications.

Multi-scale inference of interaction rules in animal groups using Bayesian model selection.

Inference of interaction rules of animals moving in groups usually relies on an analysis of large scale system behaviour. Models are tuned through repeated simulation until they match the observed behaviour. More recent work has used the fine scale motions of animals to validate and fit the rules of interaction of animals in groups. Here, we use a Bayesian methodology to compare a variety of models to the collective motion of glass prawns (Paratya australiensis). We show that these exhibit a stereotypical 'phase transition', whereby an increase in density leads to the onset of collective motion in one direction. We fit models to this data, which range from: a mean-field model where all prawns interact globally; to a spatial Markovian model where prawns are self-propelled particles influenced only by the current positions and directions of their neighbours; up to non-Markovian models where prawns have 'memory' of previous interactions, integrating their experiences over time when deciding to change behaviour. We show that the mean-field model fits the large scale behaviour of the system, but does not capture fine scale rules of interaction, which are primarily mediated by physical contact. Conversely, the Markovian self-propelled particle model captures the fine scale rules of interaction but fails to reproduce global dynamics. The most sophisticated model, the non-Markovian model, provides a good match to the data at both the fine scale and in terms of reproducing global dynamics. We conclude that prawns' movements are influenced by not just the current direction of nearby conspecifics, but also those encountered in the recent past. Given the simplicity of prawns as a study system our research suggests that self-propelled particle models of collective motion should, if they are to be realistic at multiple biological scales, include memory of previous interactions and other non-Markovian effects.

Immune challenge affects reproductive behaviour in the guppy (Poecilia reticulata).

Immunocompetence and reproduction are among the most important determinants of fitness. However, energetic and metabolic constraints create conflict between these two life-history traits. While many studies have explored the relationship between immune activity and reproductive fitness in birds and mammals inoculated with bacterial endotoxin, very few have focused on fish. Fish have been neglected in this area due, in part, to the claim that they are largely resistant to the immune effects of endotoxins. However, the present study suggests that they are susceptible to significant effects with respect to reproductive behaviour. Here, we examined the reproductive behaviour of male guppies following exposure to bacterial lipopolysaccharides (LPS) in comparison to that of male guppies in a control treatment. Additionally, we investigated the responses of females to these males. We show that although immune challenge does not suppress general activity in male guppies, it significantly reduces mating effort. While females showed no difference in general activity as a function of male treatments, they did exhibit reduced group cohesion in the presence of LPS-exposed males. We discuss this in the context of sickness behaviours, social avoidance of immune-challenged individuals and the effects of mounting an immune response on reproductive behaviour.

Thermal acclimation of interactions: differential responses to temperature change alter predator-prey relationship.

Different species respond differently to environmental change so that species interactions cannot be predicted from single-species performance curves. We tested the hypothesis that interspecific difference in the capacity for thermal acclimation modulates predator-prey interactions. Acclimation of locomotor performance in a predator (Australian bass, Macquaria novemaculeata) was qualitatively different to that of its prey (eastern mosquitofish, Gambusia holbrooki). Warm (25°C) acclimated bass made more attacks than cold (15°C) acclimated fish regardless of acute test temperatures (10-30°C), and greater frequency of attacks was associated with increased prey capture success. However, the number of attacks declined at the highest test temperature (30°C). Interestingly, escape speeds of mosquitofish during predation trials were greater than burst speeds measured in a swimming arena, whereas attack speeds of bass were lower than burst speeds. As a result, escape speeds of mosquitofish were greater at warm temperatures (25°C and 30°C) than attack speeds of bass. The decline in the number of attacks and the increase in escape speed of prey means that predation pressure decreases at high temperatures. We show that differential thermal responses affect species interactions even at temperatures that are within thermal tolerance ranges. This thermal sensitivity of predator-prey interactions can be a mechanism by which global warming affects ecological communities.

Functional coding haplotypes and machine-learning feature elimination identifies predictors of Methotrexate Response in Rheumatoid Arthritis patients.

BACKGROUND: Major challenges in large scale genetic association studies include not only the identification of causative single nucleotide polymorphisms (SNPs), but also accounting for SNP-SNP interactions. This study thus proposes a novel feature engineering approach integrating potentially functional coding haplotypes (pfcHap) with machine-learning (ML) feature selection to identify biologically meaningful, possibly causative genetic factors, that take into consideration potential SNP-SNP interactions within the pfcHap, to best predict for methotrexate (MTX) response in rheumatoid arthritis (RA) patients. METHODS: Exome sequencing from 349 RA patients were analysed, of which they were split into training and unseen test set. Inferred pfcHaps were combined with 30 non-genetic features to undergo ML recursive feature elimination with cross-validation using the training set. Predictive capacity and robustness of the selected features were assessed using six popular machine learning models through a train set cross-validation and evaluated in an unseen test set. FINDINGS: Significantly, 100 features (95 pfcHaps, 5 non-genetic factors) were identified to have good predictive performance (AUC: 0.776-0.828; Sensitivity: 0.656-0.813; Specificity: 0.684-0.868) across all six ML models in an unseen test dataset for the prediction of MTX response in RA patients. INTERPRETATION: Majority of the predictive pfcHap SNPs were predicted to be potentially functional and some of the genes in which the pfcHap resides in were identified to be associated with previously reported MTX/RA pathways. FUNDING: Singapore Ministry of Health's National Medical Research Council (NMRC) [NMRC/CBRG/0095/2015; CG12Aug17; CGAug16M012; NMRC/CG/017/2013]; National Cancer Center Research Fund and block funding Duke-NUS Medical School.; Singapore Ministry of Education Academic Research Fund Tier 2 grant MOE2019-T2-1-138.

Consensus decision making by fish.

Decisions reached through consensus are often more accurate, because they efficiently utilize the diverse information possessed by group members [1-3]. A trust in consensus decision making underlies many of our democratic political and judicial institutions [4], as well as the design of web tools such as Google, Wikipedia, and prediction markets [5, 6]. In theory, consensus for the option favored by the majority of group members will lead to improved decision-making accuracy as group size increases [2, 4]. Although group-living animals are known to utilize social information [7-10], little is known about whether or not decision accuracy increases with group size. In order to reach consensus, group members must be able to integrate the disparate information they possess. Positive feedback, resulting from copying others, can spread information quickly through the group, but it can also result in all individuals making the same, possibly incorrect, choice [8, 11, 12]. On the other hand, if individuals never copy each other, their decision making remains independent and they fail to benefit from information exchange [4]. Here, we show how small groups of sticklebacks (Gasterosteus aculeatus) reach consensus when choosing which of two replica fish to follow. As group size increases, the fish make more accurate decisions, becoming better at discriminating subtle phenotypic differences of the replicas. A simple quorum rule proves sufficient to explain our observations, suggesting that animals can make accurate decisions without the need for complicated comparison of the information they possess. Furthermore, although submission to peers can lead to occasional cascades of incorrect decisions, these can be explained as a byproduct of what is usually accurate consensus decision making.

The role of female dominance hierarchies in the mating behaviour of mosquitofish.

While studies of sexual selection focus primarily on female choice and male-male competition, males should also exert mate choice in order to maximize their reproductive success. We examined male mate choice in mosquitofish, Gambusia holbrooki, with respect to female size and female dominance. We found that the number of mating attempts made by a male was predicted by the dominance rank of females in a group, with dominant females attracting more mating attempts than subordinates. The number of mating attempts made by males was independent of the female size. The observed bias in the number of mating attempts towards dominant females may be driven either by straightforward male mate choice, since dominance and female fecundity are often closely related, or via the dominant females mediating male mating behaviour by restricting their access to subordinate females.

Social aggregation in the pelagic zone with special reference to fish and invertebrates.

Aggregations of organisms, ranging from zooplankton to whales, are an extremely common phenomenon in the pelagic zone; perhaps the best known are fish schools. Social aggregation is a special category that refers to groups that self-organize and maintain cohesion to exploit benefits such as protection from predators, and location and capture of resources more effectively and with greater energy efficiency than could a solitary individual. In this review we explore general aggregation principles, with specific reference to pelagic organisms; describe a range of new technologies either designed for studying aggregations or that could potentially be exploited for this purpose; report on the insights gained from theoretical modelling; discuss the relationship between social aggregation and ocean management; and speculate on the impact of climate change. Examples of aggregation occur in all animal phyla. Among pelagic organisms, it is possible that repeated co-occurrence of stable pairs of individuals, which has been established for some schooling fish, is the likely precursor leading to networks of social interaction and more complex social behaviour. Social network analysis has added new insights into social behaviour and allows us to dissect aggregations and to examine how the constituent individuals interact with each other. This type of analysis is well advanced in pinnipeds and cetaceans, and work on fish is progressing. Detailed three-dimensional analysis of schools has proved to be difficult, especially at sea, but there has been some progress recently. The technological aids for studying social aggregation include video and acoustics, and have benefited from advances in digitization, miniaturization, motion analysis and computing power. New techniques permit three-dimensional tracking of thousands of individual animals within a single group which has allowed novel insights to within-group interactions. Approaches using theoretical modelling of aggregations have a long history but only recently have hypotheses been tested empirically. The lack of synchrony between models and empirical data, and lack of a common framework to schooling models have hitherto hampered progress; however, recent developments in this field offer considerable promise. Further, we speculate that climate change, already having effects on ecosystems, could have dramatic effects on aggregations through its influence on species composition by altering distribution ranges, migration patterns, vertical migration, and oceanic acidity. Because most major commercial fishing targets schooling species, these changes could have important consequences for the dependent businesses.

Quorum decision-making facilitates information transfer in fish shoals.

Despite the growing interest in collective phenomena such as "swarm intelligence" and "wisdom of the crowds," little is known about the mechanisms underlying decision-making in vertebrate animal groups. How do animals use the behavior of others to make more accurate decisions, especially when it is not possible to identify which individuals possess pertinent information? One plausible answer is that individuals respond only when they see a threshold number of individuals perform a particular behavior. Here, we investigate the role of such "quorum responses" in the movement decisions of fish (three-spine stickleback, Gasterosteus aculeatus). We show that a quorum response to conspecifics can explain how sticklebacks make collective movement decisions, both in the absence and presence of a potential predation risk. Importantly our experimental work shows that a quorum response can reduce the likelihood of amplification of nonadaptive following behavior. Whereas the traveling direction of solitary fish was strongly influenced by a single replica conspecific, the replica was largely ignored by larger groups of four or eight sticklebacks under risk, and the addition of a second replica was required to exert influence on the movement decisions of such groups. Model simulations further predict that quorum responses by fish improve the accuracy and speed of their decision-making over that of independent decision-makers or those using a weak linear response. This study shows that effective and accurate information transfer in groups may be gained only through nonlinear responses of group members to each other, thus highlighting the importance of quorum decision-making.