JOINT IDENTIFICATION OF SPATIALLY VARIABLE GENES VIA A NETWORK-ASSISTED BAYESIAN REGULARIZATION APPROACH

Area of Science:

• Spatial transcriptomics
• Computational biology
• Bioinformatics

Background:

• Identifying spatially variable genes is crucial for understanding gene interactions in biological processes.
• Current methods often ignore complex gene network structures and confounding cellular composition in spatial transcriptomic data.

Purpose of the Study:

• To develop a novel Bayesian regularization approach for spatial transcriptomic data analysis.
• To effectively correct for confounding variations from cellular distributions.
• To simultaneously identify spatially variable genes and incorporate gene network structures.

Main Methods:

• A Bayesian regularization approach using thresholded graph Laplacian regularization.
• Modeling spatial transcriptomic data with a zero-inflated negative binomial distribution.
• Correction for confounding variations due to cellular heterogeneity.

Main Results:

• The proposed method effectively identifies spatially variable genes while accounting for gene networks.
• Confounding variations from cellular composition are significantly corrected.
• Demonstrated competitive performance through extensive simulations and real data applications.

Conclusions:

• The novel Bayesian method advances spatial transcriptomics analysis by integrating gene networks and correcting for cellular heterogeneity.
• This approach offers improved accuracy for identifying spatially variable genes.
• Provides a robust tool for dissecting complex biological mechanisms in a spatial context.