The screening and validation process of miR-223-3p for saliva identification.

More accurate identification of the types of body fluids left at a crime scene is indispensable for improving the judicial chain of evidence. MicroRNAs (miRNAs) have become recognized as ideal molecular markers for the identification of body fluids in forensic science due to their short length, stability and high tissue specificity. In this study, small RNA sequencing was performed on 20 samples of five types of body fluids (peripheral blood, menstrual blood, saliva, semen, and vaginal secretions) with the BGISEQ-500 sequencing platform, and the specific miRNA markers of saliva and vaginal secretions were screened by bioinformatics methods, including differential expression analysis and significant enrichment analysis. Through RT-qPCR validation of 169 samples, we confirmed that miR-223-3p can be used as a saliva-specific marker. In addition, we considered miR-223-3p in combination with four other miRNA molecules (miR-451a, miR-891a-5p, miR-144-5p, miR-203a-3p) that had been previously screened and verified in our laboratory, and seven body fluid prediction models based on machine learning algorithms were constructed and verified. The results showed that a kernel density estimation (KDE) model based on the five miRNA markers for body fluid identification could achieve 100% accuracy in the samples tested in the present study.