scMultiSim: simulation of single-cell multi-omics and spatial data guided by gene regulatory networks and cell-cell...

Area of Science:

• Computational Biology
• Genomics
• Single-cell Analysis

Background:

• Simulated single-cell data are crucial for developing and validating computational methods when experimental ground truth is unavailable.
• Existing simulators often focus on limited biological factors, hindering comprehensive benchmarking.

Purpose of the Study:

• To introduce scMultiSim, a versatile simulator for generating multimodal single-cell data.
• To provide a tool that models complex biological relationships and technical noise for realistic data simulation.

Main Methods:

• scMultiSim models gene expression, chromatin accessibility, RNA velocity, and spatial locations, capturing inter-modal relationships.
• It integrates cell identity, gene regulatory networks, cell-cell interactions, and chromatin accessibility, with adjustable parameters for technical noise.
• The simulator was used to benchmark various computational tasks, including data integration and inference methods.

Main Results:

• scMultiSim successfully generates data reflecting expected biological effects.
• Benchmarking demonstrated its utility across diverse computational tasks, such as multimodal data integration and RNA velocity estimation.
• The simulator proved effective for inferring gene regulatory and cell-cell interaction networks from spatial data.

Conclusions:

• scMultiSim offers a comprehensive platform for generating multimodal single-cell data.
• It enables robust benchmarking of a wide array of computational methods, including those for complex biological problems.
• This tool advances the development and evaluation of single-cell analysis pipelines.