This study explores informatics approaches to forensic body fluid identification via proteomic mass spectrometry.
This study introduces the use of proteomic mass spectrometry in the process of forensic body fluid identification. Body fluid identification is an important adjunct to forensic DNA analysis because it can provide contextual evidence. Protein-based methods are well suited for body fluid detection as protein markers in the three most common forensic body fluids (blood, saliva and semen) are both specific and abundant. There can, however, be inherent difficulties in identifying other forensically important body fluids such as menstrual blood, which is a “mixed” body fluid containing blood, vaginal secretions as well as fragments of endometrial tissue which are specific to menses. While endometrial marker proteins have been identified, their detection can require purification steps that increase cost and time of analysis. An alternative approach using machine learning and based on the relative abundance of both marker and non-marker proteins can be used to build an accurate predictive model. Importantly, a similar approach using regression models can be applied to the deconvolution blood, saliva and semen mixtures as well.