Ronald Koes

The University of Copenhagen invites applications for two PhD fellowships in AI-driven discovery of plant metabolic pathways, hosted by the Department of Plant and Environmental Sciences at Frederiksberg Campus. The positions are doctoral candidates DC7 and DC8 in the Marie Skłodowska-Curie Doctoral Network Stripes&Stars, which investigates modern breeding of ornamental and fruit crops through flower and fruit colour and specialized metabolites, including anthocyanins and other flavonoids in petunia, calibrachoa, poinsettia and bilberry. DC7: Knowledge graph construction for specialized-metabolite pathways This project will make fragmented scientific knowledge about plant metabolism computable. The doctoral researcher will use the PlantConnectome large-language-model pipeline to extract information from more than 70,000 publications and connect enzymes, substrates, transporters, transcription factors, stresses and genetic modifications in a comprehensive knowledge graph. The work will extend the graph to environmental contexts, apply symbolic reasoning and graph neural networks to infer missing pathway components, and develop an interactive interface for evidence-linked information on genes, pathways and phenotypes. DC8: Multi-omics co-expression networks for predicting novel metabolic pathways This project will systematically identify genes involved in plant pigment and specialized-metabolite production. The researcher will build condition-specific co-expression and regulatory networks from consortium and public RNA-seq data using the TEA-GCN algorithm, integrate cell- and tissue-resolved expression with proteomic and epigenomic datasets, and combine network evidence with the DC7 knowledge graph. Neurosymbolic reasoning and graph machine learning will be used to prioritize enzymes, regulators and environmental cues for experimental validation. Both doctoral researchers will work closely with experimental partners and contribute to a shared computational framework for predictive colour and specialized-metabolite engineering. DC7 includes secondments at TU Wien and Klemm + Sohn, while DC8 includes a secondment at the University of Amsterdam. Both projects include network-wide schools, workshops and transferable-skills training. The Mutwil group, led by Associate Professor Marek Mutwil, works at the interface of comparative and evolutionary plant biology and AI-driven biological discovery. The international group develops cross-species gene-expression atlases, comparative co-expression networks and knowledge graphs, supported by molecular biology facilities, plant-growth facilities and high-performance computing and GPU resources.

These two PhD fellowships are part of the Marie Skłodowska-Curie Doctoral Network Stripes&Stars at the Department of Plant and Environmental Sciences (PLEN), University of Copenhagen . The positions sit within an AI-driven effort to uncover plant metabolic pathways linked to flower and fruit colour, with a particular focus on specialised metabolites such as anthocyanins and other flavonoids in petunia, calibrachoa, poinsettia and bilberry. The two doctoral candidates will form the computational backbone of the network at Copenhagen and work in close connection with experimental partners across Europe. The group develops large cross-species gene expression atlases, comparative co-expression networks, knowledge graphs and machine-learning tools to convert literature and omics archives into testable biological hypotheses. The research environment combines plant biology, data science and AI, supported by modern molecular biology laboratories, plant growth facilities and high-performance computing and GPU resources on the Frederiksberg Campus. DC7 focuses on building a comprehensive knowledge graph for specialised metabolite pathways. The project will use large-language-model-based extraction to connect genes, enzymes, substrates, transporters, transcription factors, stresses and genetic modifications reported across the plant literature, then extend this resource to environmental context and reasoning-based hypothesis generation. The goal is to create a machine-readable framework that can surface missing pathway components, regulatory relationships and engineering targets, with an interactive interface for consortium use. DC8 focuses on network-based discovery from multi-omics data. This project will construct explainable co-expression and regulatory networks from RNA-seq and complementary omics resources, integrate tissue- and cell-resolved evidence, and combine these results with the DC7 knowledge graph to support linked, interpretable pathway prediction. The outputs will help prioritise candidate regulators, enzymes and environmental cues for downstream validation by the experimental partners and shared consortium resources. The doctoral candidates will also take part in network-wide secondments and training activities, including placements with partner institutions and workshops across the Doctoral Network. The principal supervisor is Associate Professor Marek Mutwil .