Chromatin accessibility, a fundamental property of DNA, plays a critical role in gene regulation and cell identity. With the rapid advances in single-cell chromatin accessibility sequencing (scCAS) technologies, the importance of accurate cell type annotation in scCAS data has become increasingly crucial. However, existing automatic annotation methods face significant limitations, including low accuracy, failure to incorporate reference data, and an inability to identify novel cell types.
Recently researchers from Nankai University developed a new approach called RAINBOW to address these challenges. RAINBOW, which is described in a Quantitative Biology paper, is a reference-guided automatic annotation method based on the contrastive learning framework, which focuses on learning common features among cells of the same type and distinguishing differences between non-similar cells. Additionally, RAINBOW incorporates information from external reference data as prior knowledge, and through unsupervised clustering, it can effectively identify novel cell types.
Comprehensive benchmarking experiments have shown that RAINBOW outperforms current leading methods in annotating both known and novel cell types. This innovative approach holds promise in uncovering new biological processes and functions of cells, further advancing our understanding of the complex chromatin regulatory landscape that controls gene transcription.