Determining the bodily origin of biological traces is a valuable tool in forensic investigations as it helps corroborate testimonies, reconstruct crime-related activities, and select relevant samples for further analysis. Current body fluid identification (BFI) methods rely on enzymatic, spectroscopic, and chemical tests, which are often limited in sensitivity and specificity. Recent research has explored novel markers for BFI, for instance metabolites, based on their potential body fluid/tissue specificity. Metabolites are small molecules produced by human and microbial cellular processes that can be measured using advanced techniques like gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS). These methodologies remain underexplored for the identification of forensically relevant body fluids/tissues. In this pilot study, we employed a high-resolution, untargeted LC-quadrupole time-of-flight (QTOF)-MS approach to investigate body fluid/tissue specific markers from nine biological fluids/tissues, including feces, fingerprick blood, menstrual blood, saliva, semen, skin from palms, urine, vaginal fluid and venous blood. We used sparse partial least-squares discriminant analysis (sPLS-DA) to identify key features responsible for body fluid/tissue-specific clustering and generalized local learning (GLL) to select features directly associated with specific body fluids/tissues. Lastly, we present nine predictive features, one for each fluid/tissue, demonstrating that our approach has the potential to be used in forensic casework.
Swayambhu et al. (2025) studied this question.