Abstract

Background: Rapid and accurate identification of Candida species at species or species-complex level is essential for antifungal stewardship and hospital infection control. An automated ITS DNA barcoding pipeline 108 TẠP CHÍ Y DƯỢC HUẾ - HUE JOURNAL OF MEDICINE AND PHARMACY ISSN 3030-4318; eISSN: 3030-4326 was developed to standardize sequence analysis, reduce processing time, and generate structured outputs suitable for laboratory information systems and infection-control surveillance. Objectives: To describe the key features of the automated ITS DNA barcoding tool and evaluate its identification performance on clinical yeast sequences using standard diagnostic metrics. Materials and methods: The pipeline was built on a quality-controlled ITS reference database covering 134 medically relevant fungal species across 7 genera and comprised four major components: K-mer indexing, K2P distance comparison with genus-specific thresholds, hierarchical three-tier reporting (species/ complex/genus), and cryptic-species flagging. The evaluation dataset comprised 145 ITS sequences: 92 from clinical yeast isolates obtained from Srinagarind Hospital, Thailand, and 53 reference sequences from out-of- database species retrieved from GenBank. Accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and processing time were calculated against phylogenetic tree-based identification as the reference standard and compared with NCBI BLAST. Results: Among 145 evaluated sequences, the tool recorded 90 true positives (TP), 52 true negatives (TN), 1 false positive (FP), and 2 false negatives (FN). Overall accuracy was 97.9%, sensitivity 97.8%, specificity 98.1%, PPV 98.9%, and NPV 96.3%. Mean runtime was 0.93 seconds/sequence. The tool achieved complete species-level concordance for C. tropicalis (24/24), C. glabrata (9/9), C. krusei (4/4), and C. africana (6/6); and safely reported results at the complex level for the C. albicans, C. parapsilosis complex, and C. rugosa complex. Compared with NCBI BLAST, the automated tool demonstrated higher accuracy (97.9% vs. 93.5%) and approximately 10-fold faster processing (0.93 vs. 10 seconds/sequence). Conclusion: The automated ITS barcoding tool demonstrated high identification accuracy, rapid processing, and safe hierarchical reporting for challenging cases. Its primary strengths are automated sequence analysis, standardized structured output, and potential utility for antifungal stewardship and hospital infection-control surveillance.