Objective: This study characterizes converging biological trends in GBA-associated Parkinson’s Disease by integrating transcriptomic profiling with exploratory image-based analysis of donor-matched midbrain organoids.

Background: High-content cellular imaging can capture disease-relevant phenotypes in Parkinson’s disease (PD) models. However, extracting meaningful insights from these images requires biological validation and an understanding of the features driving the analysis.

Method: We performed differential expression (DE) analysis on bulk RNA sequencing data from organoids derived from six donors. We used linear mixed models to account for pseudoreplication and batch effects. The DE results informed a systematic network perturbation analysis to identify dysregulated gene interactions in GBA carriers. In parallel, we used a dataset of 16,000 single-cell images from nine donors to evaluate whether automated classification models, such as DenseNets and Vision Transformers, have the potential to distinguish genotypes. We applied gradient-based attribution and attention mapping to identify the morphological regions and spatial features contributing to these classifications.

Results: The transcriptomic network analysis confirmed the dysregulation of sphingolipid signaling alongside proteostatic, senescence-associated, and broader stress-response pathways. Image-based analysis revealed that transformer-based models identified genotype-relevant information within spatially localized regions, especially near organoid boundaries and dense cellular areas. We are currently cross-referencing these spatial cues with the identified transcriptomic pathways to determine if they reflect the same underlying pathology.

Conclusion: Confounder-aware transcriptomics highlights both known and new GBA-associated regulatory genes. Preliminary image analysis suggests that specific spatial features near organoid boundaries correlate with genotype. These findings lay the groundwork for further research into how molecular dysregulation manifests as observable morphological changes in organoid models.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/integrating-transcriptomics-and-morphological-profiling-to-characterize-gba-associated-parkinsons-disease-in-midbrain-organoids/