sTPLS: identifying common and specific correlated patterns under multiple biological conditions.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Large-scale biological data offers insights into tissue development and disease.
• Existing methods struggle to identify common and condition-specific associations across diverse biological conditions.

Purpose of the Study:

• To develop a novel method for integrating multiple datasets with different biological conditions.
• To identify shared and condition-specific associations between different feature types.

Main Methods:

• Developed the sparse tensor-based partial least squares (sTPLS) method.
• Integrated pairwise datasets from different biological conditions.
• Applied sTPLS to pharmacogenomic, single-cell, and tensor-structured data.

Main Results:

• Identified condition-specific and shared gene-drug comodules across seven cancer types.
• Uncovered condition-specific and shared gene-peak comodules in single-cell data.
• Revealed shared and distinct cell communication patterns in COVID-19 patients.

Conclusions:

• sTPLS effectively identifies biologically meaningful relationships across diverse conditions.
• The method is versatile for multi-omics integrative analysis.
• sTPLS enhances understanding of tissue development and disease progression.