SpatialMETA: A Novel Framework for Integrating Spatial Transcriptomics and Metabolomics Data
- Conducted large-scale benchmarking of 7 integration methods (spaVAE, scVI, scPoli, …) across 6 multi-batch spatial transcriptomic-metabolomic datasets.
- Developed quantitative batch integration assessment system:
- Basic ST assessment (Continuity (CHAOS, PAS, …), Marker score (Moran’s, Geary’s, …), Specificity score (Specificity, Logistic regression, …)
- Biological conservation (ASW, Cell-type LISI (cLISI), …)
- Batch correction (Batch ASW, Integration LISI (iLISI), …)
- Reconstruction fidelity (Cosine Similarity, Pearson Correlation)
