sRDA

sRDA performs sparse redundancy analysis to integrate high-dimensional multi-omics data and identify predictive genetic factors and interactions that explain phenotypic variation.

Key Features:

• Integration of Multi-Omics Data: Integrates multiple omics datasets from different biological levels to capture complex phenotypic variation.
• Directional Multivariate Analysis (RDA): Employs redundancy analysis (RDA) to model directional relationships between explanatory and outcome variables and express maximal variance via linear combinations.
• Sparse Solution via Elastic Net Penalization: Incorporates Elastic Net penalization within an iterative framework to obtain sparse solutions suitable for settings where p≫n.
• High-Dimensional Data Handling: Tailored to typical omics measurements such as methylation markers and gene-expression values encountered in genomic studies.
• Predictive Variable Selection: Selects a subset of predictive variables to enhance interpretability and reduce computational complexity.

Scientific Applications:

• Genotype–Phenotype Association Analysis: Elucidates and predicts phenotypic variations by exploring genetic contributors and interactions among associated genetic factors.
• Marfan Syndrome Omics Study: Applied to a dataset of 485,512 methylation markers and 18,424 gene-expression values from 55 patients with Marfan syndrome to investigate genetic mechanisms underlying phenotypic variation.

Methodology:

An iterative redundancy analysis framework integrating Elastic Net penalization to select a subset of predictive variables; simulation studies were conducted to evaluate reliability and robustness.