REBACCA

REBACCA estimates sparse basis covariance from compositional metagenomic high-throughput sequencing data to identify significant microbial co-occurrence patterns while mitigating biases of compositional data.

Key Features:

• Compositional data handling: Tailored to metagenomic datasets and accounts for compositional constraints arising from varying sampling or sequencing depths in high-throughput sequencing.
• Regularized basis covariance estimation: Implements Regularized Estimation of Basis Covariance based on compositional data to infer covariance structure among taxa.
• Log-ratio system construction and sparse solutions: Constructs systems using log ratios of count or proportion data and targets sparse solutions for systems with deficient rank.
• l1-norm shrinkage method: Solves the constructed system using l1-norm shrinkage regularization to identify significant co-occurrence patterns and control false positives.
• Performance in sparse settings: Demonstrated higher accuracy, controlled false positive rates, and faster operation than comparable methods in simulation studies under sparse conditions.

Scientific Applications:

• Microbial co-occurrence inference: Identification of significant pairwise or network-level co-occurrence patterns among microbial taxa from metagenomic data.
• Microbial ecology and ecosystem function: Analysis of community interactions to inform studies of ecological relationships and functional dynamics in environmental microbiomes.
• Microbiome studies in health and disease: Investigation of microbial interaction patterns relevant to host-associated microbiomes and their roles in health and disease contexts.

Methodology:

Constructs systems from log ratios of count or proportion data, performs regularized estimation of basis covariance to obtain sparse solutions for deficient-rank systems, and solves the system using l1-norm shrinkage.