Standardizing a microbiome pipeline for body fluid identification from complex crime scene stains

Area of Science:

• Forensic Science
• Microbiology
• Bioinformatics
• Machine Learning

Background:

• Next-generation sequencing enables human microbiome analysis for forensic applications.
• Site-specific microbial communities can identify body fluids in biological evidence.
• Lack of standardized protocols and systematic testing hinders forensic microbiome integration.

Purpose of the Study:

• To address critical decisions in establishing standardized protocols for forensic microbiome analysis.
• To focus on bioinformatics choices and machine learning for presenting microbiome results.
• To test methods on forensically relevant and challenging samples, including mixed-source and sexome samples.

Main Methods:

• Utilized operational taxonomic units (OTUs) for read data processing.
• Developed heterogeneous training datasets for a random forest classifier.
• Incorporated six forensically relevant classes: saliva, semen, skin, urine, and vaginal/menstrual fluid.

Main Results:

• The random forest classifier achieved a high weighted average F1 score of 0.89.
• Systematic testing on mock forensic samples reliably detected components in mixtures.
• Successfully identified vaginal fluid from underwear substrates and inferred sexual activity from sexome samples.

Conclusions:

• Microbiome analysis combined with machine learning offers promising forensic applications.
• The study provides a framework for standardized protocols and systematic testing.
• The sexome (sexually shared microbiome) is valuable for assessing sexual activities in forensic investigations.