Description
- Abstract:
- Spatially resolved transcriptomics (SRT) has enabled transcriptome-wide gene expression profiling with spatial localization, leading to significant advances in un- derstanding tissue architecture and cellular interactions. As the number of com- putational methods for spatial clustering, spatially variable gene detection, and gene network analysis continues to grow, there is a critical need for high-fidelity synthetic SRT data with known ground truth to support systematic benchmarking and experimental design. However, existing simulators either lack spatial modeling or fail to preserve gene–gene co-expression, limiting their utility for comprehensive method evaluation. To address this gap, we developed SPASI, a biologically in- formed and flexible simulator for generating synthetic SRT data. SPASI supports both reference-based and reference-free simulations and uniquely preserves two essential features: spatial expression patterns across tissue regions and gene–gene correlation structures. We achieved this by integrating gene-wise count modeling with a NORTA framework, allowing for realistic simulation of count distributions and network dependencies. We evaluated SPASI using datasets from multiple SRT platforms and demonstrated that it outperformed existing simulators, in preserv- ing both gene-level and spatial-level statistical properties. SPASI maintained the lowest root-mean-square error and highest Pearson correlation to real data, and was the only method to preserve spatial patterns of key marker genes and the gene co-expression network. In reference-free mode, SPASI remained robust in re- taining both spatial and gene fidelity. Furthermore, we applied SPASI-simulated data to benchmark downstream methods, including five spatially variable gene detection tools and two co-expression network inference algorithms. Our results demonstrated SPASI’s capacity to distinguish method performance under realis- tic spatial complexity. SPASI provides a general-purpose, extensible simulation platform for spatial transcriptomics, supporting reproducible benchmarking and advancing computational method development in spatial omics.
- Notes:
- Thesis (Sc. M.)--Brown University, 2025
Citation
Fu, Jing,
"A Comprehensive Simulator for Spatially Resolved Transcriptomics"
(2025).
Biostatistics Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:qdruuwe2/
Relations
Collection:
-
Biostatistics Theses and Dissertations
Theses and Dissertations for the Biostatistics department....