NEUROSPARK

Today, drug discovery for neurodegenerative diseases follows a slow, reductionist, and trial-and-error path. NEUROSPARK aims to replace this with a predictive process based on systems biology: a Virtual Cell Platform capable of simulating cellular processes and predicting therapeutic effects by integrating chemical descriptors, omics data, imaging, aggregation assays, functional assays, and in vivo readouts. The initial focus is Spinocerebellar Ataxia Type 3 (SCA3), a rare disease with a defined genetic driver, known protein aggregation biology, mitochondrial and proteostasis dysfunction, and no disease-modifying therapies—a rigorous validation case because it combines a clear molecular pathology with already available cellular and animal models.

Objectives

• Build the Virtual Cell Platform: a multimodal AI system capable of predicting cellular perturbations and explaining the mechanisms relevant to neurodegeneration
• Generate high-quality multimodal data: a dataset compliant with FAIR principles that integrates public data, proprietary omics data, imaging, functional assays, molecular descriptors, and in silico simulations
• Validate on SCA3: calibrate the platform using SCA3-specific in vitro data and validate the predictions in a C. elegans animal model (ATX3Q130-GFP)
• Identify lead compounds: select and validate 2–3 lead compounds with measurable therapeutic benefit and a mechanistic link to the pathological pathways of SCA3