GPerturb: Gaussian process modelling of single-cell perturbation data.

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Single-cell RNA sequencing and CRISPR screening are powerful tools for analyzing genetic perturbations.
• Understanding combinatorial perturbation effects is crucial but hindered by data sparsity and complex biological mechanisms.

Purpose of the Study:

• To introduce GPerturb, a Gaussian process-based sparse perturbation regression model.
• To estimate gene-level perturbation effects and their uncertainty at single-cell resolution.

Main Methods:

• GPerturb utilizes an additive structure to distinguish signal from noise.
• The model captures sparse and interpretable effects from both discrete and continuous responses.
• It provides uncertainty estimates for perturbation effects on individual genes.

Main Results:

• GPerturb demonstrates competitive performance against state-of-the-art methods on simulated and real-world datasets.
• The model successfully reveals meaningful gene-perturbation interactions.
• Identified effects are consistent with established biological knowledge.

Conclusions:

• GPerturb offers a novel approach for analyzing complex gene expression dependencies and perturbations.
• This method advances the understanding of gene regulation at the single-cell level.