Data-Driven Modeling for Species-Level Taxonomic Assignment From 16S rRNA: Application to Human Microbiomes

Area of Science:

• Microbiology and Genomics
• Bioinformatics and Computational Biology

Background:

• Metagenomic studies commonly use 16S ribosomal RNA (16S rRNA) amplicon sequencing to estimate bacterial composition.
• High-throughput sequencing of hypervariable regions (e.g., V1-V2, V3-V4) faces challenges in distinguishing species due to sequence homology.
• Existing 16S rRNA databases show inconsistencies in bacterial lineage classification.

Purpose of the Study:

• To develop a method for enhanced species-level taxonomic assignment using 16S rRNA sequences.
• To address the challenge of distinguishing between highly homologous bacterial species.
• To improve the resolution of bacterial composition analysis in metagenomic samples.

Main Methods:

• Defined 'homologous species groups' comprising species with indistinguishable 16S rRNA sequences.
• Re-annotated bacterial lineage information across three major 16S rRNA databases based on NCBI taxonomy.
• Constructed consensus sequence models for hypervariable regions and employed a k-nearest neighbor method for taxonomic assignment.

Main Results:

• The proposed method successfully assigned species or homologous species groups, achieving maximum resolution with 16S rRNA data.
• Evaluation using simulated, mock, and real microbiome (salivary, gut) datasets showed high correlation with actual bacterial composition.
• The method enabled accurate species-level profiling and identification of bacterial composition differences between phenotypic groups.

Conclusions:

• The developed approach effectively resolves bacterial species-level taxonomy from 16S rRNA data, overcoming limitations of sequence homology.
• This method provides a robust tool for accurate metagenomic profiling and comparative microbiome studies.
• The concept of homologous species groups offers a practical solution for species identification in challenging cases.