SCiMS: Sex Calling in Metagenomic Sequences

Area of Science:

• Microbiome research
• Genomics
• Bioinformatics

Background:

• Host sex significantly influences microbial community structure but is often missing or mislabeled in studies.
• Existing sex-calling tools struggle with low host DNA concentrations, common in samples like stool.
• Accurate host sex data is crucial for understanding sex-specific biological variations.

Purpose of the Study:

• To develop a robust bioinformatic tool for accurate host sex determination in metagenomic datasets.
• To address the challenge of missing or unreliable sex metadata in microbiome research.
• To provide a scalable and generalizable solution for host sex classification across different species.

Main Methods:

• Developed SCiMS (Sex Calling in Metagenomic Sequences), a tool utilizing sex-chromosome read density ratios.
• Employed a Bayesian classifier to predict host sex from host-derived DNA in metagenomic samples.
• Validated SCiMS performance through simulations and application to real-world datasets (Human Microbiome Project, murine, chicken).

Main Results:

• SCiMS accurately predicts host sex even with minimal host DNA (e.g., >85% accuracy with ~450 host reads in simulations).
• Outperformed existing tools on Human Microbiome Project data, demonstrating superior accuracy and precision-recall balance.
• Showcased cross-species generalizability with 100% accuracy in murine and strong performance in chicken datasets.

Conclusions:

• SCiMS offers an accurate, scalable, and broadly applicable method for host sex classification in metagenomic studies.
• Enables recovery of critical sex metadata, enhancing the integrity and reliability of microbiome analyses.
• The tool serves as a vital quality control measure for microbiome research and is publicly available.