Overall objectives
The simplest ecosystem is a complex network of interactions between a diversity of organisms providing a diversity of functions. Each organism is the result of evolutionary processes driven by the creation then the selection of molecular diversity. Every function is a coordinated cascade of biochemical reactions, sensitive to substrate and environment, arising to adapt to the needs of the organism. Pleiade explores the diversity of organisms and the diversity of their functions and, as a fundamental challenge, seeks to formalize the links between them.
Pleiade measures the diversity of organisms by comparing DNA sequences and describes it using geometric methods. Amplicon sequences from metabarcoding are compared systematically to produce matrices of taxonomic distances. Mathematical analysis of these distance matrices, made possible by advances in dimension reduction, pattern recognition, and high-performance computing, reveals complex descriptions of the molecular diversity of the organisms in the sampled environment.
Pleiade examines the diversity of functions performed by these organisms by identifying the genes responsible for biochemical processes and grouping them into metabolic networks. Annotation of whole genome and metagenomic sequences allows us to delineate the functions provided by individual organisms and describe the interactions between them. Metabolic and process-based models are developed as compact descriptions of functional diversity. In addition to providing a means to simulate a system, a model is a syntactic object that symbolically represents a range of functional behaviors; patterns in the diversity of the encoded functions can be explored inferentially without exhaustively simulating the model to enumerate the set of behaviors it represents.
Comparison of annotated genomes and their associated metabolic networks reveals how functions arose over time: what functions, when they arose, and by which evolutionary mechanisms. Annotation is not performed on genomes individually, but comparatively, taking into account the similarities and differences between related species and strains.
A further challenge, developed recently, is considering the challenge of linking diversity and function in the particular context of microbial communities. We are developing a synergistic, iterative combination of a community-based strategy for deciphering the diversity in cultures and environmental samples, through metagenomic and metabolomic analysis of functional diversity and metabarcoding analysis of taxonomic diversity; and a function-based strategy for constructing digital twins of natural or designed communities through numerical models. The goal is a hybrid framework for studying systems dynamics using spatio-temporal models.
Shared methodologies needed to scale up to the complexity of biological systems, include high-performance computing (HPC); machine learning, including clustering, meta-modeling and classification for knowledge engineering; machine reasoning, specifically logical and rule-based methods used for model inference and network analysis. Logicial methods in particular promote explainable inference, since the rules are expressed in biological terms and are auditable by biologists, independently from the combinatorial and heuristic optimization techniques used to apply the rules.
Pleiade maintains strong collaborative relations with experimental biologists, and is committed to developing applications in ecology, evolution, biotechnology, and health. Team resources are dedicated to facilitating the adoption of our research by non specialist users, through development of reusable software, integration in HPC frameworks, improvement of web-based environments, and deployment of Jupyter, Galaxy, and Kubernetes interfaces.
