PStrain

PStrain infers microbial strain genotypes and abundances from shotgun metagenomic sequencing data by iteratively leveraging SNV genotype frequencies and multi-locus variant co-occurrence in MetaPhlAn2 marker genes.

Key Features:

• Iterative Profiling: Employs an iterative approach that uses genotype frequencies and co-occurrence of variants across multiple loci within the same reads to refine strain genotype and abundance estimates.
• Optimization via SNV Loci: Integrates SNV data from MetaPhlAn2 marker genes to optimize strain inference and improve both abundance estimation and genotype determination.
• Performance Improvement: Reported average increases of 87.75% in inferring strain abundances and 59.45% in determining genotypes compared to state-of-the-art methods.

Scientific Applications:

• Colorectal cancer (CRC) cohort analysis: Applied to CRC cohorts to detect strain-level differences, including distinct sequences of Bacteroides coprocola between CRC and control samples.

Methodology:

PStrain uses genotype frequencies at SNV loci and multi-locus variants covered by the same reads as evidence for single strains and iteratively applies these principles to SNVs in MetaPhlAn2 marker genes.