Longitudinal omics data analysis: approaches and applications.

Area of Science:

• Biostatistics
• Computational Biology
• Genomics

Background:

• Longitudinal omics data (LOD) is crucial for studying biological processes and disease progression over time.
• Analyzing LOD presents challenges like imbalance, high dimensionality, and non-Gaussian distributions.

Purpose of the Study:

• To review statistical and computational approaches for longitudinal omics data analysis.
• To highlight the applications and limitations of various methods for LOD.

Main Methods:

• Discussion of linear mixed models (LMM) and generalized linear mixed models (GLMM) and their extensions.
• Exploration of functional data analysis (FDA), classification, survival modeling, and multivariate methods.
• Coverage of emerging topics like data integration, clustering, and network-based modeling.

Main Results:

• Categorization of state-of-the-art approaches for omics data analysis.
• Emphasis on how different methods address specific features of longitudinal data.
• Identification of challenges and solutions in LOD modeling and hypothesis testing.

Conclusions:

• Effective analysis of complex LOD requires robust and tailored statistical strategies.
• This review provides a guideline for researchers to select appropriate methods for LOD analysis.
• Understanding LOD dynamics is key for advancing biological insights and clinical applications.