Scientific research encompasses a variety of objectives, including measurement, making predictions, identifying laws, and more. The advent of advanced measurement technologies and computational methods has largely automated the processes of big data collection and prediction. However, the discovery of laws, particularly universal ones, still heavily relies on human intellect. Even with human intelligence, complex systems present a unique challenge in discerning the laws that govern them. Even the preliminary step, system description, poses a substantial challenge. Numerous metrics have been developed, but universally applicable laws remain elusive. Due to the cognitive limitations of human comprehension, a direct understanding of big data derived from complex systems is impractical. Therefore, simplification becomes essential for identifying hidden regularities, enabling scientists to abstract observations or draw connections with existing knowledge. As a result, the concept of macrostates -- simplified, lower-dimensional representations of high-dimensional systems -- proves to be indispensable. Macrostates serve a role beyond simplification. They are integral in deciphering reusable laws for complex systems. In physics, macrostates form the foundation for constructing laws and provide building blocks for studying relationships between quantities, rather than pursuing case-by-case analysis. Therefore, the concept of macrostates facilitates the discovery of regularities across various systems. Recognizing the importance of macrostates, I propose the relational macrostate theory and a machine learning framework, MacroNet, to identify macrostates and design microstates. The relational macrostate theory defines a macrostate based on the relationships between observations, enabling the abstraction from microscopic details. In MacroNet, I propose an architecture to encode microstates into macrostates, allowing for the sampling of microstates associated with a specific macrostate. My experiments on simulated systems demonstrate the effectiveness of this theory and method in identifying macrostates such as energy. Furthermore, I apply this theory and method to a complex chemical system, analyzing oil droplets with intricate movement patterns in a Petri dish, to answer the question, ``which combinations of parameters control which behavior?'' The macrostate theory allows me to identify a two-dimensional macrostate, establish a mapping between the chemical compound and the macrostate, and decipher the relationship between oil droplet patterns and the macrostate.

With the ability to observe the atmospheres of terrestrial exoplanets via transit spectroscopy on the near-term horizon, the possibility of atmospheric biosignatures has received considerable attention in astrobiology. While traditionally exoplanet scientists looking for life focused on biologically relevant trace gases such as O2 and CH4, this approach has raised the spectre of false positives. Therefore, to address these shortcomings, a new set of methods is required to provide higher confidence in life detection. One possible approach is measuring the topology of atmospheric chemical reaction networks (CRNs). To investigate and assess this approach, the ability of network-theoretic metrics to distinguish the distance from thermochemical equilibrium in the atmosphere of hot jupiters was tested. After modeling the atmospheres of hot jupiters over a range of initial conditions using the VULCAN modeling package, atmospheric CRNs were constructed from the converged models and their topology measured using the Python package NetworkX. I found that network metrics were able to predict the distance from thermochemical equilibrium better than atmospheric species abundances alone. Building on this success, I modeled 30,000 terrestrial worlds. These models divided into two categories: Anoxic Archean Earth-like planets that varied in terms of CH4 surface flux (modeled as either biotic or abiotic in origin), and modern Earth-like planets with and without a surface flux of CCl2F2 (to represent the presence of industrial civilizations). I constructed atmospheric CRNs from the converged models, and analyzed their topology. I found that network metrics could distinguish between atmospheres with and without the presence of life or technology. In particular, mean degree and average shortest path length consistently performed better at distinguishing between abiotic and biotic Archean-like atmospheres than CH4 abundance.

Astrobiology is premised on the idea that life beyond Earth can exist. Yet, everything known about life is derivative from life on Earth. To understand life beyond Earth, then, requires a definition of life that is abstracted beyond a particular geophysical context. To do this requires a formal understanding of the physical mechanisms by which matter is animated into life. At current, such descriptions are completely lacking for the emergence of life, but do exist for the emergence of consciousness. Namely, contemporary neuroscience offers definitions for universal physical processes that are in one-to-one correspondence with conscious experience. Since consciousness is a sufficient condition for life, these universal definitions of consciousness offer an interesting way forward in terms of the search for life in the cosmos. In this work, I systematically examine Integrated Information Theory (IIT), a well-established theory of consciousness, with the aim of applying it in both biological and astrobiological settings. Surprisingly, I discover major problems with Integrated Information Theory on two fronts: mathematical and epistemological. On the mathematical side, I show how degeneracies buried deep within the theory render it mathematically ill-defined, while on the epistemological side, I prove that the postulates of IIT are scientifically unfalsifiable and inherently metaphysical. Given that IIT is the preeminent theory of consciousness in modern neuroscience, these results have far-reaching implications in this field. In addition, I show that the epistemic issues of falsifiability that hamstring IIT apply quite generally to all contemporary theories of consciousness, which suggests a major reframing of the problem is necessary. The problems that I reveal in regard to defining consciousness offer an important parallel in regard to defining life, as both fields seek to define their topic of study in absence of an existing theoretical framework. To avoid metaphysical problems related to falsifiability, universal theories of both life and consciousness must be framed with respect to independent empirical observations that can be used to benchmark predictions from the theory. In this regard, I argue that the epistemic debate over scientific theories of consciousness should be used to inform the discussion regarding theoretical definitions of life.

A swarm describes a group of interacting agents exhibiting complex collective behaviors. Higher-level behavioral patterns of the group are believed to emerge from simple low-level rules of decision making at the agent-level. With the potential application of swarms of aerial drones, underwater robots, and other multi-robot systems, there has been increasing interest in approaches for specifying complex, collective behavior for artificial swarms. Traditional methods for creating artificial multi-agent behaviors inspired by known swarms analyze the underlying dynamics and hand craft low-level control logics that constitute the emerging behaviors. Deep learning methods offered an approach to approximate the behaviors through optimization without much human intervention.

This thesis proposes a graph based neural network architecture, SwarmNet, for learning the swarming behaviors of multi-agent systems. Given observation of only the trajectories of an expert multi-agent system, the SwarmNet is able to learn sensible representations of the internal low-level interactions on top of being able to approximate the high-level behaviors and make long-term prediction of the motion of the system. Challenges in scaling the SwarmNet and graph neural networks in general are discussed in detail, along with measures to alleviate the scaling issue in generalization is proposed. Using the trained network as a control policy, it is shown that the combination of imitation learning and reinforcement learning improves the policy more efficiently. To some extent, it is shown that the low-level interactions are successfully identified and separated and that the separated functionality enables fine controlled custom training.

Mantle derived basalts along the entirety of the Earth’s Mid-Ocean Ridge (MOR) spreading centers are continuously altered by seawater, allowing the hydrosphere to subsume energy and exchange mass with the deep, slowly cooling Earth. Compositional heterogeneities inherent to these basalts—the result of innumerable geophysical and geochemical processes in the mantel and crust—generate spatial variation in the equilibrium states toward which these water-rock environments cascade. This alteration results in a unique distribution of precipitate assemblages, hydrothermal fluid chemistries, and energetic landscapes among ecosystems rooted within and above the seafloor. The equilibrium states for the full range of basalt compositional heterogeneity present today are calculated over all appropriate temperatures and extents of reaction with seawater, along with the non-equilibrium mixtures generated when hydrothermal fluids mix back into seawater. These mixes support ancient and diverse ecosystems fed not by the energy of the sun, but by the geochemical energy of the Earth. Facilitated by novel, high throughout code, this effort has yielded a high-resolution compositional database that is mapped back onto all ridge systems. By resolving the chemical and energetic consequences of basalt-seawater interaction to sub-ridge scales, alteration features that are globally homogeneous can be distinguished from those that are locally unique, guiding future field observations with testable geochemical and biochemical predictions.

Universal biology is an important astrobiological concept, specifically for the search for life beyond Earth. Over 1.2 million species have been identified on Earth, yet all life partakes in certain processes, such as homeostasis and replication. Furthermore, several aspects of biochemistry on Earth are thought to be universal, such as the use of organic macromolecules like proteins and nucleic acids. The presence of many biochemical features in empirical data, however, has never been thoroughly investigated. Moreover, the ability to generalize universal features of Earth biology to other worlds suffers from the epistemic problem of induction. Systems biology approaches offer means to quantify abstract patterns in living systems which can more readily be extended beyond Earth’s familiar planetary context. In particular, scaling laws, which characterize how a system responds to changes in size, have met with prior success in investigating universal biology.

This thesis rigorously tests the hypothesis that biochemistry is universal across life on Earth. The study collects enzyme data for annotated archaeal, bacterial, and eukaryotic genomes, in addition to metagenomes. This approach allows one to quantitatively define a biochemical system and sample across known biochemical diversity, while simultaneously exploring enzyme class scaling at both the level of both individual organisms and ecosystems. Using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the Joint Genome Institute’s Integrated Microbial Genomes and Microbiomes (JGI IMG/M) database, this thesis performs the largest comparative analysis of microbial enzyme content and biochemistry to date. In doing so, this thesis quantitatively explores the distribution of enzyme classes on Earth and adds constraints to notions of universal biochemistry on Earth.

Understanding changes and trends in biomedical knowledge is crucial for individuals, groups, and institutions as biomedicine improves people’s lives, supports national economies, and facilitates innovation. However, as knowledge changes what evidence illustrates knowledge changes? In the case of microbiome, a multi-dimensional concept from biomedicine, there are significant increases in publications, citations, funding, collaborations, and other explanatory variables or contextual factors. What is observed in the microbiome, or any historical evolution of a scientific field or scientific knowledge, is that these changes are related to changes in knowledge, but what is not understood is how to measure and track changes in knowledge. This investigation highlights how contextual factors from the language and social context of the microbiome are related to changes in the usage, meaning, and scientific knowledge on the microbiome. Two interconnected studies integrating qualitative and quantitative evidence examine the variation and change of the microbiome evidence are presented. First, the concepts microbiome, metagenome, and metabolome are compared to determine the boundaries of the microbiome concept in relation to other concepts where the conceptual boundaries have been cited as overlapping. A collection of publications for each concept or corpus is presented, with a focus on how to create, collect, curate, and analyze large data collections. This study concludes with suggestions on how to analyze biomedical concepts using a hybrid approach that combines results from the larger language context and individual words. Second, the results of a systematic review that describes the variation and change of microbiome research, funding, and knowledge are examined. A corpus of approximately 28,000 articles on the microbiome are characterized, and a spectrum of microbiome interpretations are suggested based on differences related to context. The collective results suggest the microbiome is a separate concept from the metagenome and metabolome, and the variation and change to the microbiome concept was influenced by contextual factors. These results provide insight into how concepts with extensive resources behave within biomedicine and suggest the microbiome is possibly representative of conceptual change or a preview of new dynamics within science that are expected in the future.

Biochemical reactions underlie all living processes. Their complex web of interactions is difficult to fully capture and quantify with simple mathematical objects. Applying network science to biology has advanced our understanding of the metabolisms of individual organisms and the organization of ecosystems, but has scarcely been applied to life at a planetary scale. To characterize planetary-scale biochemistry, I constructed biochemical networks using global databases of annotated genomes and metagenomes, and biochemical reactions. I uncover scaling laws governing biochemical diversity and network structure shared across levels of organization from individuals to ecosystems, to the biosphere as a whole. Comparing real biochemical reaction networks to random reaction networks reveals the observed biological scaling is not a product of chemistry alone, but instead emerges due to the particular structure of selected reactions commonly participating in living processes. I perform distinguishability tests across properties of individual and ecosystem-level biochemical networks to determine whether or not they share common structure, indicative of common generative mechanisms across levels. My results indicate there is no sharp transition in the organization of biochemistry across distinct levels of the biological hierarchy—a result that holds across different network projections.

Finally, I leverage these large biochemical datasets, in conjunction with planetary observations and computational tools, to provide a methodological foundation for the quantitative assessment of biology’s viability amongst other geospheres. Investigating a case study of alkaliphilic prokaryotes in the context of Enceladus, I find that the chemical compounds observed on Enceladus thus far would be insufficient to allow even these extremophiles to produce the compounds necessary to sustain a viable metabolism. The environmental precursors required by these organisms provides a reference for the compounds which should be prioritized for detection in future planetary exploration missions. The results of this framework have further consequences in the context of planetary protection, and hint that forward contamination may prove infeasible without meticulous intent. Taken together these results point to a deeper level of organization in biochemical networks than what has been understood so far, and suggests the existence of common organizing principles operating across different levels of biology and planetary chemistry.

The origin of Life on Earth is the greatest unsolved mystery in the history of science. In spite of progress in almost every scientific endeavor, we still have no clear theory, model, or framework to understand the processes that led to the emergence of life on Earth. Understanding such a processes would provide key insights into astrobiology, planetary science, geochemistry, evolutionary biology, physics, and philosophy. To date, most research on the origin of life has focused on characterizing and synthesizing the molecular building blocks of living systems. This bottom-up approach assumes that living systems are characterized by their component parts, however many of the essential features of life are system level properties which only manifest in the collective behavior of many components. In order to make progress towards solving the origin of life new modeling techniques are needed. In this dissertation I review historical approaches to modeling the origin of life. I proceed to elaborate on new approaches to understanding biology that are derived from statistical physics and prioritize the collective properties of living systems rather than the component parts. In order to study these collective properties of living systems, I develop computational models of chemical systems. Using these computational models I characterize several system level processes which have important implications for understanding the origin of life on Earth. First, I investigate a model of molecular replicators and demonstrate the existence of a phase transition which occurs dynamically in replicating systems. I characterize the properties of the phase transition and argue that living systems can be understood as a non-equilibrium state of matter with unique dynamical properties. Then I develop a model of molecular assembly based on a ribonucleic acid (RNA) system, which has been characterized in laboratory experiments. Using this model I demonstrate how the energetic properties of hydrogen bonding dictate the population level dynamics of that RNA system. Finally I return to a model of replication in which replicators are strongly coupled to their environment. I demonstrate that this dynamic coupling results in qualitatively different evolutionary dynamics than those expected in static environments. A key difference is that when environmental coupling is included, evolutionary processes do not select a single replicating species but rather a dynamically stable community which consists of many species. Finally, I conclude with a discussion of how these computational models can inform future research on the origins of life.

What makes living systems different than non-living ones? Unfortunately this question is impossible to answer, at least currently. Instead, we must face computationally tangible questions based on our current understanding of physics, computation, information, and biology. Yet we have few insights into how living systems might quantifiably differ from their non-living counterparts, as in a mathematical foundation to explain away our observations of biological evolution, emergence, innovation, and organization. The development of a theory of living systems, if at all possible, demands a mathematical understanding of how data generated by complex biological systems changes over time. In addition, this theory ought to be broad enough as to not be constrained to an Earth-based biochemistry. In this dissertation, the philosophy of studying living systems from the perspective of traditional physics is first explored as a motivating discussion for subsequent research. Traditionally, we have often thought of the physical world from a bottom-up approach: things happening on a smaller scale aggregate into things happening on a larger scale. In addition, the laws of physics are generally considered static over time. Research suggests that biological evolution may follow dynamic laws that (at least in part) change as a function of the state of the system. Of the three featured research projects, cellular automata (CA) are used as a model to study certain aspects of living systems in two of them. These aspects include self-reference, open-ended evolution, local physical universality, subjectivity, and information processing. Open-ended evolution and local physical universality are attributed to the vast amount of innovation observed throughout biological evolution. Biological systems may distinguish themselves in terms of information processing and storage, not outside the theory of computation. The final research project concretely explores real-world phenomenon by means of mapping dominance hierarchies in the evolution of video game strategies. Though the main question of how life differs from non-life remains unanswered, the mechanisms behind open-ended evolution and physical universality are revealed.