Monozygotic twins (MZTs) have highly homologous nuclear genomes and cannot be effectively differentiated by traditional forensic Short Tandem Repeat(STR)markers, which remains a persistent challenge in forensic individual identification. A previous study had shown that Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) diversity in the human microbiota affords excellent discriminatory power in forensic individual identification, with performance superior to that of 16S rRNA gene-based profiling. However, whether such diversity differs between MZTs has not been adequately studied. Therefore, our study aimed to evaluate whether CRISPR diversity in the human salivary microbiome can be used for the forensic discrimination of MZTs as a polymorphic marker. For this purpose, saliva samples at five time points (days 0, 7, 14, 21, 30) from 11 twin pairs (9 MZT, 2 dizygotic twins) were collected. Amplicon sequencing targeted two CRISPR loci from oral streptococci (CRISPR1 from S. equinus, and CRISPR2 from S. thermophilus) and the 16S rRNA gene were performed. The results showed that the CRISPR1/2 loci exhibited a unique ASV profile specific to each individual. Among MZTs possessing almost identical nuclear genetic profiles, the majority of ASVs were unique to each individual, with shared ASVs constituting only 2.3% to 56.6% of the total ASVs. PCoA based on Bray-Curtis distance clustered longitudinal samples from the same individual, and nearly all MZT pairs were clearly separated (R²> 0.3, p < 0.05). Inter-individual differences in CRISPR diversity between MZTs, DZTs, and unrelated individuals were consistently high and significantly greater than intra-individual temporal variations. The CRISPR1/2 markers showed good longitudinal stability and reliable discrimination ability within 30 days (ICC > 0.6, CV < 10%, Intra-group Kruskal-Wallis test, p＞0.05 ). Compared to the 16S rRNA gene, CRISPR1/2 demonstrated greater ability to differentiate MZTs and needed less sequencing depth. In binary pairwise MZT discrimination, CRISPR1/2 achieved a mean AUC > 0.9 and high prediction accuracy via LASSO Logistic, random forest, and XGBoost models (94.4%–100% mean accuracy). Our findings suggest that the diversity of salivary CRISPR reliably differentiate between MZT pairs, making it a promising tool for resolving difficult individual identification cases in future forensic work.

Authors

• Tingyi Liang (Shanxi Medical University, China)
• Jiaxuan Hao (Shanxi Medical University, China)
• Agu Damu (Shanxi Medical University, China)
• Xinyu Li (Shanxi Medical University, China)
• Yanyu Lu (Shanxi Medical University, China)
• Haohui Wang (Shanxi Medical University, China)
• Xue Han (Shanxi Medical University, China)
• Kaige Wang (Shanxi Medical University, China)
• Daijing Yu (Shanxi Medical University, China)
• Liuyaoxing Zhang (Shanxi Medical University, China)
• Niu Gao (Shanxi Medical University, China)
• Wenjing Hu (Shanxi Medical University, China)
• Jun Zhang (Shanxi Medical University, China)
• Jiangwei Yan (Shanxi Medical University, China)