跨批次单细胞空间组学数据多粒度聚类方法及系统
By employing a triplet alignment method based on a dual-path parallel graph encoder and online pseudo-label generation, the bottlenecks of multi-scale clustering and cross-batch integration in existing spatial omics methods are overcome. This enables self-driven discovery and robust integration of multi-granularity cellular and tissue structures, improving the reliability of cellular heterogeneity characterization and data integration.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG UNIV
- Filing Date
- 2026-05-08
- Publication Date
- 2026-07-17
AI Technical Summary
Existing spatial omics methods cannot simultaneously reveal macroscopic organizational structure and fine-grained cellular niches in cluster analysis, and fine-grained information is lost during cross-batch integration, resulting in excessive smoothing of biologically significant cellular heterogeneity.
A dual-path parallel graph encoder is used to model molecular expression and spatial structure separately. Combined with online pseudo-label generation and spatial topology-aware triplet alignment, unsupervised discovery and robust cross-batch integration of multi-granularity cellular tissue structures are achieved.
Without external annotation, we achieved multi-scale clustering discovery from coarse-grained tissue domains to fine-grained cellular niches, preserving fine-grained biological heterogeneity within samples and improving the accuracy of cellular heterogeneity characterization and the reliability of cross-batch data integration.
Smart Images
Figure CN122135793B_ABST