Robust additive models for high dimensional biological data with interaction effects
Author(s): Tahiru Mahama
Abstract: High-dimensional biological datasets such as gene expression profiles, epigenomic markers, and multi-omics measurements are central to deciphering complex biological systems. However, their analysis is challenged by dimensionality, noise, and interaction effects. Classical linear models often underperform in such settings, and while generalized additive models (GAMs) offer flexibility for nonlinear modeling, they struggle with scalability, robustness, and interaction capture in high-dimensional contexts. This study reviews recent advancements in robust additive models tailored for high-throughput biological data. These models incorporate sparsity-promoting regularization, robust estimation techniques, and structured interaction modeling, thereby enhancing inference reliability and interpretability. Theoretical foundations and computational strategies are discussed in depth. We highlight applications across genomics, transcriptomics, and systems biology, demonstrating the effectiveness of robust additive models in capturing complex biological relationships. Emerging approaches such as Sparse Additive Models (SpAM), Neural Additive Models (NAMs), and hierarchical interaction structures are evaluated for their capacity to model gene to gene and gene environment interactions. Persistent challenges include handling data heterogeneity, missingness, and integrating multiple omics layers. Future directions emphasize the development of scalable, interpretable, and biologically adaptive models to meet the growing demands of precision medicine and functional genomics.
DOI: 10.22271/multi.2023.v5.i8a.740Pages: 24-28 | Views: 662 | Downloads: 157Download Full Article: Click Here
How to cite this article:
Tahiru Mahama.
Robust additive models for high dimensional biological data with interaction effects. Int J Multidiscip Trends 2023;5(8):24-28. DOI:
10.22271/multi.2023.v5.i8a.740
Research Journals