Dealing with dimensionality: the application of machine learning to multi-omics data.

MOTIVATION: Machine learning (ML) methods are motivated by the need to automate information extraction from large datasets in order to support human users in data-driven tasks. This is an attractive approach for integrative joint analysis of vast amounts of omics data produced in next generation sequencing and other -omics assays. A systematic assessment of the current literature can help to identify key trends and potential gaps in methodology and applications. We surveyed the literature on ML multi-omic data integration and quantitatively explored the goals, techniques and data involved in this field. We were particularly interested in examining how researchers use ML to deal with the volume and complexity of these datasets. RESULTS: Our main finding is that the methods used are those that address the challenges of datasets with few samples and many features. Dimensionality reduction methods are used to reduce the feature count alongside models that can also appropriately handle relatively few samples. Popular techniques include autoencoders, random forests and support vector machines. We also found that the field is heavily influenced by the use of The Cancer Genome Atlas dataset, which is accessible and contains many diverse experiments. AVAILABILITY AND IMPLEMENTATION: All data and processing scripts are available at this GitLab repository: https://gitlab.com/polavieja_lab/ml_multi-omics_review/ or in Zenodo: https://doi.org/10.5281/zenodo.7361807. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Endosymbiosis before eukaryotes: mitochondrial establishment in protoeukaryotes.

Endosymbiosis and organellogenesis are virtually unknown among prokaryotes. The single presumed example is the endosymbiogenetic origin of mitochondria, which is hidden behind the event horizon of the last eukaryotic common ancestor. While eukaryotes are monophyletic, it is unlikely that during billions of years, there were no other prokaryote-prokaryote endosymbioses as symbiosis is extremely common among prokaryotes, e.g., in biofilms. Therefore, it is even more precarious to draw conclusions about potentially existing (or once existing) prokaryotic endosymbioses based on a single example. It is yet unknown if the bacterial endosymbiont was captured by a prokaryote or by a (proto-)eukaryote, and if the process of internalization was parasitic infection, slow engulfment, or phagocytosis. In this review, we accordingly explore multiple mechanisms and processes that could drive the evolution of unicellular microbial symbioses with a special attention to prokaryote-prokaryote interactions and to the mitochondrion, possibly the single prokaryotic endosymbiosis that turned out to be a major evolutionary transition. We investigate the ecology and evolutionary stability of inter-species microbial interactions based on dependence, physical proximity, cost-benefit budget, and the types of benefits, investments, and controls. We identify challenges that had to be conquered for the mitochondrial host to establish a stable eukaryotic lineage. Any assumption about the initial interaction of the mitochondrial ancestor and its contemporary host based solely on their modern relationship is rather perilous. As a result, we warn against assuming an initial mutually beneficial interaction based on modern mitochondria-host cooperation. This assumption is twice fallacious: (i) endosymbioses are known to evolve from exploitative interactions and (ii) cooperativity does not necessarily lead to stable mutualism. We point out that the lack of evidence so far on the evolution of endosymbiosis from mutual syntrophy supports the idea that mitochondria emerged from an exploitative (parasitic or phagotrophic) interaction rather than from syntrophy.

Efficient assembly and long-term stability of defensive microbiomes via private resources and community bistability.

Understanding the mechanisms that promote the assembly and maintenance of host-beneficial microbiomes is an open problem. Empirical evidence supports the idea that animal and plant hosts can combine 'private resources' with the ecological phenomenon known as 'community bistability' to favour some microbial strains over others. We briefly review evidence showing that hosts can: (i) protect the growth of beneficial strains in an isolated habitat, (ii) use antibiotics to suppress non-beneficial, competitor strains, and (iii) provide resources that only beneficial strains are able to translate into an increased rate of growth, reproduction, or antibiotic production. We then demonstrate in a spatially explicit, individual-based model that these three mechanisms act similarly by selectively promoting the initial proliferation of preferred strains, that is, by acting as a private resource. The faster early growth of preferred strains, combined with the phenomenon of 'community bistability,' allows those strains to continue to dominate the microbiome even after the private resource is withdrawn or made public. This is because after a beneficial colony reaches a sufficiently large size, it can resist invasion by parasites without further private support from the host. We further explicitly model localized microbial interactions and diffusion dynamics, and we show that an intermediate level of antibiotic diffusion is the most efficient mechanism in promoting preferred strains and that there is a wide range of parameters under which hosts can promote the assembly of a self-sustaining defensive microbiome. This in turn supports the idea that hosts readily evolve to promote host-beneficial defensive microbiomes.

Analysis of stability to cheaters in models of antibiotic degrading microbial communities.

Antibiotic resistance carried out by antibiotic degradation has been suggested recently as a new mechanism to maintain coexistence of microbial species competing on a single limiting resource, even in well-mixed homogeneous environments. Species diversity and community stability, however, critically depend on resistance against social cheaters, mutants that do not invest in production, but still enjoy the benefits provided by others. Here we investigate how different mutant cheaters affect the stability of antibiotic producing and degrading microbial communities. We consider two cheater types, production and degradation cheaters. We generalize the mixed inhibition-zone and chemostat models introduced previously [Kelsic, E. D., Zhao, J., Vetsigian, K., Kishony, R., 2015. Counteraction of an tibiotic production and degradation stabilizes microbial communities. Nature521, 516-519.] to study the population dynamics of microbial communities in well-mixed environment, and analyze the invasion of different cheaters in these models. We show that production cheaters, mutants that cease producing antibiotics, always destroy coexistence whenever there is a cost of producing these antibiotics. Degradation cheaters, mutants that loose their function of producing extracellular antibiotic degrading molecules, induce community collapse only if the cost of producing the degradation factors is above a critical level. Our analytical studies, supported by numerical simulations, highlight the sensitivity of antibiotic producing and degrading communities to loss-of-function mutants.

Evolution of the division of labor between genes and enzymes in the RNA world.

The RNA world is a very likely interim stage of the evolution after the first replicators and before the advent of the genetic code and translated proteins. Ribozymes are known to be able to catalyze many reaction types, including cofactor-aided metabolic transformations. In a metabolically complex RNA world, early division of labor between genes and enzymes could have evolved, where the ribozymes would have been transcribed from the genes more often than the other way round, benefiting the encapsulating cells through this dosage effect. Here we show, by computer simulations of protocells harboring unlinked RNA replicators, that the origin of replicational asymmetry producing more ribozymes from a gene template than gene strands from a ribozyme template is feasible and robust. Enzymatic activities of the two modeled ribozymes are in trade-off with their replication rates, and the relative replication rates compared to those of complementary strands are evolvable traits of the ribozymes. The degree of trade-off is shown to have the strongest effect in favor of the division of labor. Although some asymmetry between gene and enzymatic strands could have evolved even in earlier, surface-bound systems, the shown mechanism in protocells seems inevitable and under strong positive selection. This could have preadapted the genetic system for transcription after the subsequent origin of chromosomes and DNA.

Strategy diversity stabilizes mutualism through investment cycles, phase polymorphism, and spatial bubbles.

There is continuing interest in understanding factors that facilitate the evolution and stability of cooperation within and between species. Such interactions will often involve plasticity in investment behavior, in response to the interacting partner's investments. Our aim here is to investigate the evolution and stability of reciprocal investment behavior in interspecific interactions, a key phenomenon strongly supported by experimental observations. In particular, we present a comprehensive analysis of a continuous reciprocal investment game between mutualists, both in well-mixed and spatially structured populations, and we demonstrate a series of novel mechanisms for maintaining interspecific mutualism. We demonstrate that mutualistic partners invariably follow investment cycles, during which mutualism first increases, before both partners eventually reduce their investments to zero, so that these cycles always conclude with full defection. We show that the key mechanism for stabilizing mutualism is phase polymorphism along the investment cycle. Although mutualistic partners perpetually change their strategies, the community-level distribution of investment levels becomes stationary. In spatially structured populations, the maintenance of polymorphism is further facilitated by dynamic mosaic structures, in which mutualistic partners form expanding and collapsing spatial bubbles or clusters. Additionally, we reveal strategy-diversity thresholds, both for well-mixed and spatially structured mutualistic communities, and discuss factors for meeting these thresholds, and thus maintaining mutualism. Our results demonstrate that interspecific mutualism, when considered as plastic investment behavior, can be unstable, and, in agreement with empirical observations, may involve a polymorphism of investment levels, varying both in space and in time. Identifying the mechanisms maintaining such polymorphism, and hence mutualism in natural communities, provides a significant step towards understanding the coevolution and population dynamics of mutualistic interactions.

Beneficial laggards: multilevel selection, cooperative polymorphism and division of labour in threshold public good games.

BACKGROUND: The origin and stability of cooperation is a hot topic in social and behavioural sciences. A complicated conundrum exists as defectors have an advantage over cooperators, whenever cooperation is costly so consequently, not cooperating pays off. In addition, the discovery that humans and some animal populations, such as lions, are polymorphic, where cooperators and defectors stably live together--while defectors are not being punished--, is even more puzzling. Here we offer a novel explanation based on a Threshold Public Good Game (PGG) that includes the interaction of individual and group level selection, where individuals can contribute to multiple collective actions, in our model group hunting and group defense. RESULTS: Our results show that there are polymorphic equilibria in Threshold PGGs; that multi-level selection does not select for the most cooperators per group but selects those close to the optimum number of cooperators (in terms of the Threshold PGG). In particular for medium cost values division of labour evolves within the group with regard to the two types of cooperative actions (hunting vs. defense). Moreover we show evidence that spatial population structure promotes cooperation in multiple PGGs. We also demonstrate that these results apply for a wide range of non-linear benefit function types. CONCLUSIONS: We demonstrate that cooperation can be stable in Threshold PGG, even when the proportion of so called free riders is high in the population. A fundamentally new mechanism is proposed how laggards, individuals that have a high tendency to defect during one specific group action can actually contribute to the fitness of the group, by playing part in an optimal resource allocation in Threshold Public Good Games. In general, our results show that acknowledging a multilevel selection process will open up novel explanations for collective actions.

Cooperators Unite! Assortative linking promotes cooperation particularly for medium sized associations.

BACKGROUND: Evolution of cooperative behaviour is widely studied in different models where interaction is heterogeneous, although static among individuals. However, in nature individuals can often recognize each other and chose, besides to cooperate or not, to preferentially associate with or to avoid certain individuals.Here we consider a dynamical interaction graph, in contrast to a static one. We propose several rules of rejecting unwanted partners and seeking out new ones, and study the probability of emergence and maintenance of cooperation on these dynamic networks. RESULTS: Our simulations reveal that cooperation can evolve and be stable in the population if we introduce preferential linking, even if defectors can perform it too. The fixation of cooperation has higher probability than that of on static graphs, and this effect is more prevalent at high benefit to cost ratios. We also find an optimal number of partners, for which the fixation probability of cooperation shows a maximum. CONCLUSIONS: The ability to recognize, seek out or avoid interaction partners based on the outcome of past interactions has an important effect on the emergence of cooperation. Observations about the number of partners in natural cooperating groups are in concordance with the result of our model.

The dynamics of the RNA world: insights and challenges.

The RNA world hypothesis of the origin of life, in which RNA emerged as both enzyme and information carrier, is receiving solid experimental support. The prebiotic synthesis of biomolecules, the catalytic aid offered by mineral surfaces, and the vast enzymatic repertoire of ribozymes are only pieces of the origin of life puzzle; the full picture can only emerge if the pieces fit together by either following from one another or coexisting with each other. Here, we review the theory of the origin, maintenance, and enhancement of the RNA world as an evolving population of dynamical systems. The dynamical view of the origin of life allows us to pinpoint the missing and the not fitting pieces: (1) How can the first self-replicating ribozyme emerge in the absence of template-directed information replication? (2) How can nucleotide replicators avoid competitive exclusion despite utilizing the very same resources (nucleobases)? (3) How can the information catastrophe be avoided? (4) How can enough genes integrate into a cohesive system in order to transition to a cellular stage? (5) How can the way information is stored and metabolic complexity coevolve to pave to road leading out of the RNA world to the present protein-DNA world?