POSTm

POSTm implements a phylogeny-guided OTU-specific association test (POST) that incorporates evolutionary relationships among microbial taxa to improve detection of associations between operational taxonomic units (OTUs) and host health outcomes.

Key Features:

• Phylogenetic Information Integration: Leverages phylogenetic data in a supervised manner to adaptively borrow information from neighboring OTUs based on phylogenetic distance and observed outcome–OTU associations.
• Kernel Machine Framework: Implements POST within a kernel machine framework to model complex OTU-level effects and extend beyond traditional community-level microbiome tests.
• Adaptive Information Borrowing: Dynamically determines the extent of information borrowing from phylogenetically related taxa to improve signal detection while minimizing false positives.
• Performance Enhancement: Demonstrates improved power in simulations when phylogenetic trees are informative and maintains comparable performance when phylogenetic signal is weak.

Scientific Applications:

• Bacterial Vaginosis and Preterm Birth Studies: Applied to studies of bacterial vaginosis and preterm birth to identify biologically relevant OTUs associated with these conditions.
• Improved Detection Accuracy: Enhances selection performance of microbiome features by improving true-positive and false-positive detection rates through adaptive phylogenetic borrowing.

Methodology:

POSTm uses a local collapsing test approach that integrates phylogenetic information into OTU-specific association testing within a kernel machine framework, employing supervised adaptive borrowing based on phylogenetic distance and observed outcome–OTU associations.