Hi, I'm running SpatialCellChat on a multi-sample Visium dataset for comparison analysis and have a question about how identifyOverExpressedGenes spatially variable gene detection handles multiple samples.
When using selection.method = "meringue" or "moransi", the function uses object@images$coordinates directly to build the spatial neighbour graph. In a multi-sample object, I assume samples share the same coordinate space since each sample's tissue positions start from roughly the same pixel origin.
- How does
identifyOverExpressedGenes account for this multi-sample coordinate overlap problem?
- Is there a recommended workflow for multi-sample spatial SVG detection ? Would adding a large coordinate offset per sample before creating the object (e.g. 1000 px, which exceeds the maximum interaction range threshold) be a valid workaround?
I appreciate your feedback on this.
Hi, I'm running SpatialCellChat on a multi-sample Visium dataset for comparison analysis and have a question about how
identifyOverExpressedGenesspatially variable gene detection handles multiple samples.When using
selection.method = "meringue" or "moransi", the function usesobject@images$coordinatesdirectly to build the spatial neighbour graph. In a multi-sample object, I assume samples share the same coordinate space since each sample's tissue positions start from roughly the same pixel origin.identifyOverExpressedGenesaccount for this multi-sample coordinate overlap problem?I appreciate your feedback on this.