StrainSpy

StrainSpy enables strain-level microbiome-wide association studies from metagenomic ANI profiles.

Key Features:

• Joint strain variation and replacement detection: Captures both strain variations and strain replacement across longitudinal and multi-cohort studies without predefined identity thresholds.
• ANI-based analysis: Uses k-mer-based Average Nucleotide Identity (ANI) estimates as an orthogonal metric for strain-level analysis, avoiding spurious correlations from compositional abundance data and microbial load.
• High sensitivity: Achieved greater than 98% sensitivity in simulations versus 79% for comparable approaches at similar specificity.
• Machine learning compatibility: Selected features support predictive models that match abundance-based accuracy while avoiding microbial load confounding.

Scientific Applications:

• Longitudinal antibiotic response studies: Revealed reduced strain-level diversity despite species persistence and showed previously reported E. coli associations were driven by reduced microbial load.
• Colorectal cancer multi-cohort analysis: Identified novel strain-level associations in E. coli with adhesin gene enrichment across 3,414 metagenomes from 17 cohorts, while recovering established associations.
• Strain-level microbiome-disease association: Enables detection of strain-level associations that species-level abundance approaches miss due to overlooking intra-species genetic variation.

Methodology:

StrainSpy is a statistical algorithm that jointly models strain variations and replacement. It uses k-mer-based ANI estimates from metagenomes as input and operates without predefined identity thresholds.