Exploring the correspondence between gene expression and thalamic nuclei using the THALMANAC resource

Meghan A. Turner, Thomas Chartrand, Mathew T. Summers, Marcus Hooper, Cindy T. J. van Velthoven, Jack Waters, Saskia de Vries, Hongkui Zeng, Bosiljka Tasic, Karel Svoboda, Brian LongbioRxiv 2025 · DOI

Abstract

The thalamus connects the sensory organs and major subcortical brain regions with the neocortex. The thalamus has long been divided into multiple discrete nuclei, based on cytoarchitecture, histochemical stains, and mesoscale connectivity. However, thalamic nuclei do not completely describe thalamic organization. For example, some boundaries between thalamic nuclei are disputed, whereas other nuclei are known to contain subdomains with distinct connectivity and function. Moreover, the correspondence between cellular gene expression and other properties of thalamic projection neurons remains to be established. Spatial analysis of single cell gene expression provides a basis for reevaluating thalamic organization. We present the THALMANAC, (THALamus MERFISH ANalysis and ACcess) a Findable, Accessible, Interoperable, Reusable and Reproducible (FAIRR) resource for exploring and analyzing single-cell transcriptomic variation in the thalamus. The THALMANAC provides streamlined access to thalamic gene expression data registered to the common coordinate framework and tools for quantitative analysis and visualization of these data, all encapsulated in a reproducible, cloud computing platform. Using this resource, we find that gene expression generally supports the parcellation of thalamus into distinct nuclei. Some nuclei, such as the anteromedial nucleus, are additionally composed of discrete subdomains, while other nuclei share patterns of gene expression or are arrayed on a spatial gradient of gene expression. The THALMANAC establishes spatial transcriptomic data as a foundation for delineating thalamic organization.

Introduction

The thalamus serves as the brain's primary relay station, connecting sensory organs and subcortical regions to the neocortex. Traditionally, it has been partitioned into discrete nuclei based on cell shape, density, and connectivity. However, these anatomical boundaries are frequently contested across different reference atlases, and many nuclei are known to harbor functional subdomains. To address the need for a high-resolution molecular map, we developed the THALMANAC (THALamus MERFISH ANalysis and ACcess), a FAIRR resource for exploring single-cell transcriptomic variation in the thalamus.

Overview of the THALMANAC resource, integrating ABC Atlas data with spatial analysis tools.

Methods

The THALMANAC platform integrates several key datasets and analytical tools to provide a unified molecular map of the thalamus:

  • Spatial Transcriptomics: We utilized MERFISH data from the Allen Brain Cell (ABC) Atlas, profiling 500 genes across 80,170 individual neurons in the mouse thalamus.
  • Coordinate Mapping: Cells were registered to the Allen Common Coordinate Framework (CCFv3) to enable direct comparison with classical anatomical boundaries.
  • Analytical Framework: We employed unsupervised methods, including SpaGCN for discrete spatial domain detection and Non-negative Spatial Factorization (NSF) for identifying continuous gene expression gradients.
  • Reproducible Infrastructure: The resource is hosted on Code Ocean, featuring interactive Streamlit apps for no-code exploration, Jupyter notebooks for deep-dive analysis, and Cirrocumulus for data visualization.
The THALMANAC ecosystem provides interactive tools for no-code exploration and deep-dive analysis.

Results

Our analysis reveals that while gene expression patterns largely validate traditional thalamic parcellations, they also uncover previously hidden organizational features.

Molecular Subdomains and Gradients

Some nuclei, such as the anteromedial (AM) nucleus, exhibit discrete internal subdomains defined by specific marker genes like Pdyn and Shisa6. Conversely, other regions are defined by transcriptomic gradients rather than sharp boundaries; for instance, the transition between the ventro-anterior-lateral (VAL) and ventromedial (VM) nuclei is characterized by a gradual shift in cell populations.

Transcriptomic Diversity

Using the Inverse Simpson’s Index (ISI), we quantified the molecular diversity of different regions. The Subparafascicular area (SPA) emerged as the most diverse, containing 29 distinct clusters, whereas the Anterodorsal (AD) and Ventral Posteromedial (VPM) nuclei were found to be highly homogeneous. Unsupervised validation via SpaGCN identified 30 spatial domains that showed strong overlap with classical nuclei, such as the Reticular nucleus (RT).

Shared Molecular Logic

Interestingly, somatosensory nuclei (VPM, VPL, and PO) share several transcriptomic types, suggesting a common molecular "toolkit" for processing distinct sensory modalities.

Discussion and Conclusion

The THALMANAC establishes spatial transcriptomics as a foundational tool for delineating thalamic organization. By providing a "no-code" interface alongside raw data and reproducible code, it lowers the barrier for researchers to validate findings or identify gene markers for new experiments.

Future work will expand this resource to account for biological sex differences and individual variability. Furthermore, integrating this transcriptomic map with connectomics and physiological data will be essential for creating a truly comprehensive atlas of thalamic function. We envision THALMANAC as a living resource, evolving through community-driven annotations to refine our understanding of the brain's central hub.

Summary updated

Abstract

The thalamus connects the sensory organs and major subcortical brain regions with the neocortex. The thalamus has long been divided into multiple discrete nuclei, based on cytoarchitecture, histochemical stains, and mesoscale connectivity. However, thalamic nuclei do not completely describe thalamic organization. For example, some boundaries between thalamic nuclei are disputed, whereas other nuclei are known to contain subdomains with distinct connectivity and function. Moreover, the correspondence between cellular gene expression and other properties of thalamic projection neurons remains to be established. Spatial analysis of single cell gene expression provides a basis for reevaluating thalamic organization. We present the THALMANAC, (THALamus MERFISH ANalysis and ACcess) a Findable, Accessible, Interoperable, Reusable and Reproducible (FAIRR) resource for exploring and analyzing single-cell transcriptomic variation in the thalamus. The THALMANAC provides streamlined access to thalamic gene expression data registered to the common coordinate framework and tools for quantitative analysis and visualization of these data, all encapsulated in a reproducible, cloud computing platform. Using this resource, we find that gene expression generally supports the parcellation of thalamus into distinct nuclei. Some nuclei, such as the anteromedial nucleus, are additionally composed of discrete subdomains, while other nuclei share patterns of gene expression or are arrayed on a spatial gradient of gene expression. The THALMANAC establishes spatial transcriptomic data as a foundation for delineating thalamic organization.
BibTeX
@misc{turner2025exploring,
  author = {Meghan A. Turner and Thomas Chartrand and Mathew T. Summers and Marcus Hooper and Cindy T. J. van Velthoven and Jack Waters and Saskia de Vries and Hongkui Zeng and Bosiljka Tasic and Karel Svoboda and Brian Long},
  title = {{Exploring the correspondence between gene expression and thalamic nuclei using the THALMANAC resource}},
  year = {2025},
  doi = {10.1101/2025.09.30.679413},
  howpublished = {bioRxiv},
}
Figure