scMultiSim: simulation of multi-modality single cell data guided by cell-cell interactions and gene regulatory...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Simulated single-cell data is crucial for developing and validating computational methods when experimental ground truth is unavailable.
• Current single-cell simulators often model limited biological factors, restricting their ability to capture real-world data complexity and multi-modality.

Purpose of the Study:

• To introduce scMultiSim, an advanced in silico simulator for generating multi-modal single-cell data.
• To enable the simulation of gene expression, chromatin accessibility, RNA velocity, and spatial locations, capturing inter-modal relationships.

Main Methods:

• scMultiSim jointly models biological factors: cell identity, gene regulatory networks (GRNs), cell-cell interactions (CCIs), and chromatin accessibility.
• The simulator incorporates technical noise and allows adjustable influence of each biological factor.
• It generates multi-modal single-cell data, including spatial information.

Main Results:

• Validated scMultiSim's ability to simulate realistic biological effects.
• Demonstrated scMultiSim's utility in benchmarking diverse computational tasks, including clustering, trajectory inference, and multi-modal data integration.
• Showcased applications in RNA velocity, GRN inference, and cell-cell interaction inference from spatial transcriptomics data.

Conclusions:

• scMultiSim offers a comprehensive platform for simulating complex, multi-modal single-cell data.
• It surpasses existing simulators in its capacity to benchmark a wider array of computational problems and potential future tasks.