Redundancy effects on spread and control in network dynamics: applications in computational biomedicine

As the current COVID-19 pandemic demonstrates, our social lives and overall public health depend heavily on interactions that cross scales from the molecular networks of minute pathogens to all our transportation, health, economy, ecology, and governance networks. The study of networks that quickly link the tiniest virus to the most potent economy has been pursued by the interdisciplinary field of complex networks& systems (CNS), which aims to find and simplify commonalities among networks that are measured and studied at various scales in the natural and social sciences. Networks can be cast formally as graphs, hypergraphs, simplicial complexes, relations, or any type of deterministic or stochastic multivariate dynamical system. This formal flexibility makes CNS a potent tool in better understanding and managing multiscale network problems such as pandemics. CNS is closely correlated with Artificial Intelligence (AI), computer science, and data science, since its general-purpose methods hinge on massive combinatorial searches and inferences from big data. The close integration of CNS, AI, and computational science has become increasingly needed and pursued given the recent availability of large amounts of data about human behavior at different scales of organization, ranging from the molecular factors of disease to the collective behavior of the brain or society [T1-5]. The interdisciplinary, computational agenda of CNS has already been successfully translated to solve problems in social science, public health, physics, biology, neuroscience, and in a long and growing list of other fruitful examples [1, 2]. Indeed, the most useful, interpretable decision scenarios for COVID19 transmission have been produced by CNS scientists using massive data processing and machine learning methods. They have produced actionable models of the multiscale network interactions involved in this problem, from the molecular to human mobility and trade [3]. Indeed, pandemics such as this have long been long predicted in CNS [4]. While most of advances in CNS have come from the study of patterns of connectivity (network structure), which provides many insights into the organization of complex systems, a critical gap remains in understanding how the structure of networks affects the dynamics of complex systems [1, 2]. For instance, in brain networks we do not know how synaptic connectivity leads to the dynamical patterns of functional connectivity that are responsible for human behavior [5]. Likewise, while we know much about the network patterns of gene and protein regulation from existing systems biology models [6, 7], we have shown that the structure of interactions from these models is not sufficient to predict regulatory dynamics or derive control strategies that allow us, for instance, to revert a diseased cell to a healthy state [T1-2][8, 9]. Similar issues arise with the large-scale collection of social behavior data from social media and mobile devices, which has sparked much interest in the CNS community [T3]. The structure of social interaction can help us understand health and disease, e.g. the spread of pandemics and detection of drug interactions [T3][10, 11], but attention to dynamical processes and actions on those networks is required for us to be able to predict and control biomedical phenomena. Indeed, the COVID-19 pandemic showed that the countries that best dealt with containing the pandemic before vaccines were available (e.g. Taiwan, Hong Kong, and S. Korea) relied on precise information about both the structure (friendship and family links) and dynamics (individual and collective movement) of social networks via cell phone, credit card transactions, and even social media behavior data [12]. Our project addresses the critical gap with an original insight: in addition to patterns of connectivity and patterns of dynamics, there are important patterns of redundancy which dictate how structure affects dynamics in networks. The research we propose leverages computational CNS approaches to produce both a novel theoretical account of redundancy in complex systems, as well as computational methods to predict the most important pathways that control network dynamics in models of biochemical regulation, brain function, psychopathology, and epidemic spread. The promise and duty of sound interdisciplinary research is to work across various fields without trivializing any of them. Thus, we pursue tasks in synergistic collaboration with domain-specific scientists in four computational biomedicine problem areas: systems biology, network neuroscience, network psychopathology, and digital epidemiology. Moreover, our projectis a unique opportunity to foster Portuguese CNS leadership and training with a much-needed interdisciplinary translation of computational/data science to applications that are more than ever very important for individual and collective human health.