Akademik Veri Yönetim Sistemi
International Cumhuriyet Articial Intelligence Applications Conference 2025 (CAIAC'25), Sivas, Türkiye, 25 - 26 Eylül 2025, ss.21-27, (Tam Metin Bildiri)
Single-cell RNA sequencing (scRNA-seq) enables high-resolution characterization of cellular heterogeneity, offering unprecedented insights into complex biological systems. A key step in translating these data into biological knowledge is accurate cell-type annotation, which defines the unique transcriptomic profiles of individual cells. However, manual annotation is often impractical in large-scale single-cell studies due to its reliance on expert knowledge, time-intensive nature, and inherent subjectivity. Machine learning-based annotation provides a powerful alternative, allowing the automated extraction of meaningful biological insights from scRNA-seq data. This study provides a comparative evaluation of supervised machine learning methods for cell-type annotation in scRNA-seq data. We assessed the performance of widely used algorithms across multiple datasets representing diverse tissue types, sample sizes, and cellular compositions. Our results show that the k-Nearest Neighbors (kNN) and Random Forest (RF) classifiers achieved the most balanced trade-off between accuracy and sensitivity. These findings underscore the potential of automated scRNA-seq analysis to enhance the accuracy, reproducibility, and efficiency of cell-type annotation in largescale single-cell studies.