Fig. 6: scGALA Enhances Spatial Transcriptomics Imputation via Alignment-guided Graph Reconstruction for Multi-omics Integration.

a Imputation accuracy across datasets: UMAPs compare ground truth (14,630 genes), simulated sparse data (500 high-variance genes), scGALA-imputed data (1,545 genes; using alignment-guided graph reconstruction), and imputed-only genes (1,050 genes). Adjusted Rand Index (ARI) quantifies clustering concordance with ground truth cell types. b Functional fidelity of imputed data: Gene Ontology Biological Process (GOBP) enrichment analysis comparing scGALA-imputed dataset and simulated sparse dataset showing (\(-{\log }_{10}(\,{\mbox{FDR}})\)) values across cell types. The number before term name indicates cell types. Non-significant terms are labeled as “ns''. P-values calculated using one-sided Student’s (t)-test. c Cell type-specific marker gene recovery: Dot plots show Pearson’s correlation between imputed and ground truth expression for key marker genes, validating scGALA's precision in preserving biological signals. P-value is obtained from two-sided Pearson correlation test. d Spatial domain preservation: Spatial clustering using GraphST on ground truth and scGALA-imputed data (retaining original spatial coordinates). ARI quantifies agreement between scGALA-imputed spatial domains and ground truth annotations (clustering resolution optimized for ground truth labels). Source data are provided as a Source Data file.