Statistical analysis of spatially resolved transcriptomic data by incorporating multiomics auxiliary information

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Multiplicity problems in statistical analysis necessitate effective false discovery rate (FDR) control.
• Spatial transcriptomic data analysis benefits from incorporating external information to increase statistical power.
• Existing methods often lack the ability to integrate multiple diverse external datasets.

Purpose of the Study:

• To extend the OrderShapeEM procedure for incorporating multiple external omics studies.
• To enhance the detection of spatial expression patterns in high-dimensional spatial transcriptomic data.
• To boost statistical power while maintaining robust FDR control.

Main Methods:

• Utilized spatial pattern recognition via kernels for primary spatial transcriptomic data analysis.
• Constructed auxiliary covariates by combining information from multiple external omics studies (e.g., bulk and single-cell RNA-seq) using the Cauchy combination rule.
• Extended and implemented the OrderShapeEM method for covariate-assisted multiple testing with integrated multiomics data.

Main Results:

• The extended OrderShapeEM method demonstrated substantial power gain in detecting genes with spatial expression patterns.
• Simulations with known ground truth confirmed the method's performance.
• Case studies in various tissues showed significant improvements over classic approaches lacking external data integration.

Conclusions:

• The integrative analysis method effectively leverages multiomics data to improve spatial expression pattern detection.
• This approach offers a powerful tool for analyzing complex spatial transcriptomic datasets.
• The enhanced OrderShapeEM method provides a robust framework for covariate-assisted multiple testing in multi-omics integration.