OmicsLonDA

OmicsLonDA identifies time intervals where omics features show significant differences between groups in longitudinal studies.

Key Features:

• Semi-parametric statistical framework: Implements a semi-parametric framework to analyze longitudinal omics data.
• Smoothing splines: Uses smoothing splines to model temporal trajectories of features.
• Permutation-based inference: Constructs an empirical distribution via a permutation procedure to infer significance of temporal intervals.
• Temporal interval detection: Identifies specific time intervals where omics features are differentially regulated between groups.
• Multi-omics and physiological support: Applies to proteomics, lipidomics, metabolomics, transcriptomics, microbiome profiles, and physiological parameters from wearable sensors.
• Robustness to longitudinal data issues: Handles nonuniform sampling intervals, missing data points, subject dropout, and varying numbers of samples per subject.
• Benchmarking performance: Evaluated on five simulated datasets with diverse temporal patterns, achieving specificity >0.99 and sensitivity >0.72.

Scientific Applications:

• Integrative Personal Omics Profiling (iPOP) cohort: Revealed differential regulation of amino acids, lipids, and hormone metabolites between male and female subjects following a respiratory infection.
• Longitudinal multi-omics in pregnancy (preeclampsia): Identified potential lipid markers that are temporally significant between pregnant women with and without preeclampsia.
• Simulated dataset evaluation: Used five simulated datasets with diverse temporal patterns to benchmark detection sensitivity and specificity.

Methodology:

Applies a semi-parametric statistical framework using smoothing splines and a permutation procedure to construct an empirical distribution for inferring significant temporal intervals.