Ridge-penalized adaptive Mantel test and its application in imaging genetics

Area of Science:

• Neuroscience
• Genetics
• Statistical Science

Background:

• High-dimensional data analysis is crucial in fields like imaging genetics.
• Existing methods may not effectively capture complex associations between diverse feature sets.
• Understanding the interplay between genetic factors and brain function is a key research area.

Purpose of the Study:

• To introduce a novel statistical method, the ridge-penalized adaptive Mantel test (AdaMant), for assessing associations between two high-dimensional feature sets.
• To theoretically demonstrate how ridge penalization unifies different distance metrics and linear models in association testing.
• To apply AdaMant in a real-world imaging genetics study.

Main Methods:

• Development of the adaptive Mantel test (AdaMant) incorporating a ridge penalty.
• Theoretical analysis of ridge penalization's role in bridging Euclidean and Mahalanobis distances.
• Application of AdaMant to analyze associations between electroencephalogram (EEG) coherence and genetic data.

Main Results:

• AdaMant effectively tests associations across multiple metrics simultaneously.
• Ridge penalization provides a unified framework for association measurement and testing.
• The study identified significant associations between brain connectivity and genetic features in healthy adults.

Conclusions:

• AdaMant offers a powerful tool for high-dimensional association studies, particularly in imaging genetics.
• The method has theoretical implications for penalized hypothesis testing.
• The findings contribute to understanding the genetic underpinnings of visual working memory.