Gene-set integrative analysis of multi-omics data using tensor-based association test.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Technological advances enable multi-omics data generation across various platforms.
• Analyzing multi-omics data for novel disease gene discovery presents high-dimensional inference challenges.
• Current regression-based methods struggle with complex relationships in multi-omics datasets.

Purpose of the Study:

• To develop an efficient tensor-based framework for variable-wise inference in multi-omics analysis.
• To improve the discovery of novel disease genes and biological mechanisms.
• To enhance the modeling efficiency and computational performance of multi-omics data analysis.

Main Methods:

• Introduced a tensor-based framework to account for the matrix structure of individual multi-omics data.
• Incorporated relationships among omics effects to reduce parameters and boost modeling efficiency.
• Derived a variable-specific tensor test and enhanced computational efficiency for tensor modeling.

Main Results:

• The tensor-based framework demonstrated favorable performance compared to baseline methods in simulations.
• Applied the method to Cancer Cell Line Encyclopedia (CCLE) data, yielding valuable biological insights.
• The proposed tensor methods effectively integrate multi-omics data for robust inference.

Conclusions:

• The developed tensor-based framework offers a powerful approach for multi-omics data analysis.
• This method facilitates the discovery of novel disease genes and mechanisms.
• The framework enhances biological insights and modeling efficiency in complex biological data.