I am writing to explore a potential collaboration bridging advanced computational prediction with your team's state-of-the-art experimental validation.

At the Bioinformatics group of Forschungszentrum Jülich (IBG-4), we are developing crop.IAEGER, an AI-driven platform for the predictive design of plant traits. To navigate strict regulatory frameworks and consumer skepticism regarding transgenics, our goal is to shift from disruptive genome modifications to minimally invasive regulatory tuning.

We have built strong foundational models—deepCRE and deepCIS—that predict gene expression and transcription factor binding directly from DNA sequence. However, we are highly aware of the inherent limitations of purely in silico predictions. To advance this platform from a computational tool to a reliable engine for the bioeconomy, we critically need robust in vivo validation in crops.

We are looking for experimental partners to help us close the knowledge gap between computational target prediction and actual biological effect. A collaboration could focus on:

Experimental Validation & Platform Refinement: We need your expertise in advanced mutagenesis and base-editing (e.g., in tomato or prototyping in Marchantia) to test our AI-generated regulatory targets. Measuring the real-world effects of these predicted SNPs will be vital for refining our models and understanding pleiotropic effects on gene regulatory networks.

Targeting Agronomic Traits in Crops: We aim to apply our genRE ensemble modeling tool directly to trait-related genes, specifically focusing on identifying minimal, highly precise gain- or loss-of-function edits in key regulators (like MYB, bZIP, ERF/DREB) to optimize complex crop traits.

Integrating Broader DNA-Based Models: We view crop-IAEGER as an adaptable ecosystem. We are highly interested in synergizing our current deep learning omics framework with other DNA-based modeling approaches—such as emerging large language models (LLMs) or structural genomic models—to further enhance prediction accuracy and overcome current platform limitations.

Ultimately, crop-IAEGER provides the computational heavy lifting to predict minimal regulatory edits, but we need your experimental proficiency to validate, de-risk, and realize these consumer-friendly traits in actual crops.

best regards,
Dr. Simon M. Zumkeller

The Forschungszentrum Jülich operates as a limited liability company, officially registered as Forschungszentrum Jülich GmbH. It was initially established as a registered association (e.V.) but transitioned to its current GmbH status in 1967.
The center's research focuses heavily on solving grand societal challenges across three primary domains: Energy, Information, and Sustainable Bioeconomy. This includes developing new applications from biological resources and making agriculture fit for climate change.

The computational backbone of FZJ is managed by the Jülich Supercomputing Centre (JSC), which operates some of the most powerful supercomputing and data infrastructures in the world. Amongst, others leading infrastructure, FZJ is the host site for JUPITER, Europe's first exascale supercomputer. It is designed to surpass one quintillion calculations per second (1 ExaFLOP/s) and provides unmatched performance for artificial intelligence research and large-scale simulations.

The Szymanski Lab, officially recognized for its work in "Systems Biology and Machine Learning for Better Crops," is led by Dr. Jędrzej Jakub Szymański. The lab operates at the cutting edge of computational plant biology through a prominent dual affiliation at two of Germany's leading research institutions. The team utilizes deep learning models trained on extensive multi-species sequence and omics datasets to decipher the cis-regulatory code and protein-DNA interactions. Their highly accurate tools (such as deepCRE and deepCIS) reconstruct gene networks, assess how genetic variation impacts phenotypes, and ultimately guide targeted gene editing for precise expression modulation. By integrating multi-modal data—including genomic, transcriptomic, metabolomic, and phenomic information—the lab pinpoints exactly how genetic variations and metabolites interact during plant development and environmental stress responses.