scSpatialSIM: a simulator of spatial single-cell molecular data

Area of Science:

• Spatial biology
• Computational biology
• Bioinformatics

Background:

• Advancements in spatial molecular technologies like multiplex immunofluorescence (mIF) and spatial transcriptomics (SRT) necessitate robust statistical methods for tissue spatial architecture analysis.
• A lack of standardized "gold standard" approaches hinders effective benchmarking and comparison of these analytical methods.

Purpose of the Study:

• To develop "scSpatialSIM", an R package for simulating biologically realistic spatial single-cell molecular data.
• To provide a flexible framework for generating diverse spatial data without requiring reference datasets, enabling efficient simulation of cell clustering, co-localization, and tissue features.

Main Methods:

• Developed "scSpatialSIM", an R package for simulating spatial single-cell molecular data.
• Incorporated features for cell clustering, co-localization, tissue compartments, and tissue holes.
• Supported simulation of categorical and continuous data, integrating with existing R packages for downstream analysis.

Main Results:

• Applied "scSpatialSIM" to benchmark spatial point pattern summary functions (Ripley's K(r), G(r), g(r)).
• Ripley's K(r) demonstrated consistent detection of clustering across various radii, showing superior sensitivity and robustness compared to other methods.
• The package effectively simulates cell clustering and co-localization patterns.

Conclusions:

• "scSpatialSIM" offers a flexible and scalable platform for generating spatial data, facilitating the comparative evaluation of spatial statistics.
• The package supports the development of novel methods for characterizing tissue spatial organization and advancing spatial molecular research.
• Enables researchers to explore hypothetical scenarios and gain insights into tissue architecture and cellular interactions.