The abstract is read at the publisher; the summary is JClub's.
Spectrast enables multiscale spatial autocorrelation for single-cell spatial transcriptomics by computing convolution-based statistics on rasterized gene expression using FFT, scaling to hundreds-of-thousands of cells.
- Why it matters: Existing spatial statistics struggle with imaging-based datasets that are very large in cell count and gene number, limiting analysis of spatial organization and condition-to-condition reorganization.
- What they did: The toolkit reformulates standard spatial-autocorrelation measures as convolutions on a rasterized gene expression matrix and evaluates them with the fast Fourier transform across multiple length scales for every gene.
- The result: It makes multiscale spatial statistics tractable per gene and extends to differential reorganization testing between two tissue conditions.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.