LongDat

LongDat performs covariate-sensitive longitudinal analysis in R to distinguish direct and indirect effects of interventions and to identify mechanistic intermediate covariates in high-dimensional datasets.

Key Features:

• R package: Implemented as an R package for analysis of longitudinal multivariable data.
• Covariate-sensitive analysis: Handles a large number of covariates simultaneously for longitudinal studies.
• Differentiation of effects: Distinguishes between direct and indirect effects of treatments or interventions.
• Identification of mechanistic intermediates: Pinpoints covariates that may serve as mechanistic intermediaries.
• Data type versatility: Applicable to binary, categorical, and continuous datasets.
• Longitudinal microbiome focus: Tailored for analysis of longitudinal microbiome data.
• Computational efficiency: Employs computational techniques aimed at high accuracy, efficiency, and low memory usage.
• Benchmarking: Tested against MaAsLin2, ANCOM, lgpr, and ZIBR using simulated and real datasets with reported improvements in accuracy, runtime, and memory cost for multi-covariate datasets.

Scientific Applications:

• Longitudinal microbiome analysis: Assess temporal changes and treatment effects in microbiome studies.
• Multivariable longitudinal studies: Dissect direct and indirect treatment effects in high-dimensional longitudinal data.
• Biomarker identification: Support robust searches for biomarkers in complex longitudinal datasets.
• Analysis across data types: Apply to longitudinal studies containing binary, categorical, and continuous measurements.

Methodology:

LongDat employs computational techniques targeting high accuracy, efficiency, and low memory usage and was benchmarked against MaAsLin2, ANCOM, lgpr, and ZIBR using simulated and real datasets, with comparative analyses reporting superior accuracy, runtime, and memory cost for datasets containing multiple covariates.