scDesign2: a transparent simulator that generates high-fidelity single-cell gene expression count data with gene...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Benchmarking experimental protocols and computational methods in single-cell transcriptomics is crucial.
• Existing computational simulators struggle to simultaneously preserve genes, capture gene correlations, and generate data with variable sequencing depths.

Purpose of the Study:

• To introduce scDesign2, a novel computational simulator for single-cell transcriptomics.
• To address the limitations of current simulators in generating high-fidelity synthetic data.

Main Methods:

• scDesign2 employs transparent probabilistic models.
• Gene correlations are captured using copulas.
• The simulator generates synthetic single-cell gene expression count data.

Main Results:

• scDesign2 successfully preserves genes during simulation.
• It accurately captures complex gene correlations.
• The simulator generates high-fidelity synthetic data for multiple single-cell technologies with varying cell numbers and sequencing depths.

Conclusions:

• scDesign2 offers a transparent and effective solution for generating realistic synthetic single-cell transcriptomic data.
• This advancement aids in benchmarking experimental protocols and computational methods in the field.