A global × global test for testing associations between two large sets of variables.

Area of Science:

• Genomics
• Biostatistics
• Bioinformatics

Background:

• High-dimensional omics studies generate multiple molecular profiles per patient.
• Identifying complex associations, like copy number-regulated expression in pathways, is crucial.
• Existing methods often focus on univariate responses, limiting analysis of multivariate data.

Purpose of the Study:

• To present a novel approach for testing associations between two sets of variables in high-dimensional omics data.
• To generalize the global test for high-dimensional multivariate responses.
• To enable the detection of complex biological relationships within molecular data.

Main Methods:

• Developed a novel multivariate global test (G2) to assess associations between variable sets.
• Generalized the concept of the global test to accommodate high-dimensional multivariate responses.
• Applied the G2 method to simulated and publicly available omics datasets.

Main Results:

• The multivariate global test (G2) was compared against the univariate global test.
• Performance evaluation was conducted on simulated and real-world omics datasets.
• The G2 method demonstrated its capability in detecting multivariate associations.

Conclusions:

• The novel multivariate global test (G2) effectively identifies complex associations in high-dimensional omics data.
• This approach extends the utility of global tests for multivariate biological data analysis.
• The method is implemented in R and will be available in the globaltest package.