Human-Centered Health AI

This project aims to understand how evolution shapes animal physiology in response to prolonged exposure to juvenile undernutrition, whereby juvenile animals are forced to grow and develop in spite of chronic nutrient shortage. It will address four general questions:(1) In what way can evolution modify metabolism of a growing juvenile animal to alleviate consequences of undernutrition for Darwinian fitness?(2) What changes in allocation of metabolic resources does this adaptation entail?(3) To what degree are these changes mediated by genetic variants in 'master' genes with large effects on multiple aspects of the adaptation?(4) To what degree is evolution of tolerance to poor diet mediated by genetic variants in genes that regulate growth (i.e., the 'demand' for biomass building blocks) versus genes whose products catalyze and regulate their acquisition and allocation (i.e., the 'supply' side of growth)? The project will use lines of Drosophila melanogaster characterized by extraordinary genetically-based tolerance to larval undernutrition, a unique resource generated through >17 years of laboratory experimental evolution. To address the above questions, we will elucidate causal changes in growth regulation and metabolism underlying this highly polygenic and phenotypically complex evolutionary adaptation. First, we will quantify the rate of amino acid turnover and the allocation of resources to different components of biomass (proteins, triglycerides, glycogen, etc.) in the 'Selected' (malnutrition-tolerant) and 'Control' (unselected) larvae. Together with gene expression data, these data will be used to model the core metabolic network of Drosophila. This will generate testable predictions about differences in metabolic fluxes underlying malnutrition tolerance and identify nodes of the metabolic network critical for differential fluxes and resulting allocation patterns (question 1 and 2). It will also be used to evaluate the roles of nutrient supply versus demand on metabolic output in shaping the patterns of metabolic flux (question 4).Second, we will verify the contribution of a cis-regulatory variant in an ecdysone oxidase gene fiz to enhanced tolerance to undernutrition, and test whether the adaptive value of this variant is contingent on the presence of other elements of this complex adaptation. fiz emerged as a candidate with potential large effects on this adaptation (questions 3); it is thought to regulate growth by deactivating ecdysone, but this hypothesis appears incompatible with the direction of its effects on growth. To elucidate the effect of variation in fiz on ecdysteroid signaling we will study its effect on the abundance of different ecdysteroid species, identifying those most likely to mediate its growth-regulating effect. Third, we will combine the two above threads by investigating how fiz expression affects the rate of nutrient acquisition, metabolite abundance and the allocation of metabolic resources. Given that fiz is thought to act by modulating the demand for key metabolites, we will be able to ascertain to what degree increased demand for biomass building blocks has effects that propagate through metabolism and affect resource allocation question 4). To achieve these aims, we will combine experimental evolution with state-of-the art approaches, including genome editing, metabolomics, isotope tracing, high resolution mass spectroscopy and genome-scale flux balance analysis. This project will advance our understanding of a poorly understood and ecologically important evolutionary adaptation. It will also throw light on the broader fundamental question of how changes indifferent metabolic and regulatory elements interact to generate complex adaptations that enhance Darwinian fitness under conditions of environmental stress. Through novel application of recent experimental techniques and system-scale computational models the project will expand the boundaries of dissecting this complexity.

MetaboLinkAI aspires to revolutionize the analysis and interpretation of metabolomics data through a multidisciplinary approach that combines a comprehensive knowledge graph hub (MetaKH) with cutting-edge artificial intelligence (AI) and machine learning (ML) techniques. The project's main goals are to enhance the querying and ease of use of metabolomics data, improve research efficiency, and stimulate creativity in the field. These objectives are set to surpass current standards by creating an encyclopedic and expandable knowledge base, integrating advanced AI to handle the uncertainties of experimental data, and enabling a broader range of hypothesis testing and evaluation. The research approach is structured around three interconnected pillars:WP1 - Metabolomics Use Cases: To ensure that the project tackles relevant use cases, we chose two significant and challenging areas in which metabolomics is key: the analysis of metabolites in a biomedical context, i.e. to infer metabolic activity and regulation, and the analysis of metabolite in natural product research, i.e. for characterization of chemodiversity and bioactive scaffold discovery. These use cases are designed to guide the project's development and provide benchmarks for its progress.WP2 - Knowledge Representation and Management: The creation of an open knowledge hub, MetaKH, is central to the project. It involves (i) aggregating existing resources on chemicals, ontologies, reactions and pathways, bioactivity, publications, etc., (ii) enriching this foundational knowledge graph with metabolomics data, and (iii) establishing federated querying capabilities over the integrated knowledge hub. WP3 - AI Research Assistant and Graph Machine Learning: This pillar focuses on developing innovative methodologies and tools, such as natural language processing and graph mining methods, to enhance data interaction, analysis capabilities, and representation of uncertainty. An AI research assistant will facilitate direct interaction with the data and knowledge through querying and summarizing, while integrated graph mining methods will address the ontological characteristics of metabolomics data.MetaboLinkAI aims to create significant shifts in metabolomics research by democratizing data mining, broadening hypothesis evaluation, and transforming education in the field. The integration of AI technologies aims to address data uncertainties, enhance explainability, and facilitate complex data analysis. Looking ahead, the project envisages expanding its infrastructure to incorporate other omics data, laying the groundwork for a holistic, AI-augmented discovery platform in life sciences research.

Cancer is one of the leading causes of death worldwide and in Valais, imposing a profound burden on patients, families, and the healthcare system. Altered cell metabolism is a hallmark of cancer; thus, understanding how cancer cells reprogram their metabolism is crucial for precision oncology and the successful treatment of cancer patients. However, current molecular measurements alone cannot fully explain how tumors reprogram molecular pathways to grow, survive, or resist treatment. RNA profiles are already used in clinical settings, for example, to classify tumor subtypes or to guide therapeutic decisions, e.g., through gene expression signatures. These approaches treat RNA as a descriptive snapshot. Here, we propose to take a step further by using RNA to infer a complete picture of cellular metabolism, that is, the intricate network of chemical reactions that sustain cell’s growth and survival. Genome-scale metabolic models (GEMs) provide a static, mechanistic map of cellular metabolism that RNA profiles can constrain. However, most existing approaches simply overlay expression data onto a metabolic model for a single sample at a time; they do not learn how metabolism behaves across diverse cancer types and conditions. As a result, today’s methods cannot capture the underlying regulatory processes that shape metabolic activity.

Our project introduces a fundamentally new approach. Inspired by how large language models learn patterns from vast text corpora, we propose a self-supervised learning framework that harnesses large collections of RNA-seq profiles to uncover the hidden relationships between gene expression and metabolism. The system learns directly from data, without requiring annotated examples, allowing it to capture regulatory behaviors that are not explicitly encoded in current models. Once trained, the framework can generalize across many cancer contexts and specialize to an individual patient, producing personalized metabolic profiles that may support treatment planning, subtype identification, and the discovery of metabolic vulnerabilities.

The objective of this project is to transform RNA-seq data into a functional and predictive description of cancer cell metabolism, enabling the identification of metabolic vulnerabilities relevant to precision oncology. To achieve this, we aim to learn the hidden regulatory processes underlying the network of metabolic reactions. Specifically, the project pursues the following aims: (1) to systematically assemble and curate large-scale RNA-seq datasets relevant to cancer metabolism; (2) to develop a deep learning surrogate of genome-scale metabolic models that efficiently captures metabolic behaviour; (3) to design a cross-validation framework that optimizes coherence between transcriptomic data and metabolic predictions, forming the basis of a self-supervised learning strategy; (4) to extend the surrogate into a predictive model capable of learning hidden regulatory processes that control metabolic activity; and (5) to validate the complete framework on established cancer benchmarks, with the longer-term goal of enabling patient-specific metabolic characterization to support personalized cancer treatment.

Our proposition represents a major innovation and has the potential to transform RNA-seq data (already widely available in oncology) into physiological insights, enabling scalable, automated, and clinically actionable characterization of cancer metabolism. Embedded within the strong and expanding health ecosystem of Valais and aligned with strategic priorities in AI at both the cantonal and institutional levels, the proposed project will also nucleate new, important partnerships between HES-SO, Idiap, and the hospital. Beyond its direct impact, it will also reinforce the canton’s position as a leader in AI and health innovation.