Research map / Graph learning / Graph self-supervised learning
When Genes Speak: A Semantic-Guided Framework for Spatially Resolved Transcriptomics Data Clustering
Jiangkai Long, Yanran Zhu, Chang Tang, Kun Sun, Yuanyuan Liu, Xuesong Yan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LLM-derived gene semantic embeddings (ablation) | SemST · DLPFC 151509 | ARI: 67.37 -> 73.90 | Improved |
| add LLM-derived gene semantic embeddings (ablation) | SemST · MBA | ARI: 44.83 -> 53.92 | Improved |
| use biologically relevant LLM embeddings instead of random embeddings (ablation) | SemST · DLPFC 151509 | ARI: 59.95 -> 73.90 | Improved |
| replace biological LLM embeddings with random embeddings (versus no semantics) (ablation) | SemST · DLPFC 151509 | ARI: 67.37 -> 59.95 | Hurt |
| use LLM embeddings instead of plain BERT encoder (ablation) | SemST · DLPFC 151509 | ARI: 63.27 -> 73.90 | Improved |
| use FSM instead of cross-attention fusion (ablation) | SemST · DLPFC 151509 | ARI: 62.40 -> 73.90 | Improved |
| use FSM instead of concatenation fusion (ablation) | SemST · MBA | ARI: 46.64 -> 53.92 | Improved |
| use FSM instead of addition fusion (ablation) | SemST · DLPFC 151671 | ARI: 74.14 -> 78.70 | Improved |
| add FSM module to STAGATE | STAGATE · HBC | ARI: 44.66 -> 58.16 | Improved |
| add FSM module to Spatial-MGCN | Spatial-MGCN · HBC | NMI: 70.83 -> 69.44 | Hurt |
| add FSM module to STAGATE | STAGATE · MBA | NMI: 72.49 -> 72.11 | Hurt |
| add FSM module to baselines overall | nine baseline methods · HBC and MBA | ARI | Improved |
Closest papers by meaning
- Global Context-aware Representation Learning for Spatially Resolved Transcriptomics (ICML 2025 · similarity 0.89)
- SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics (ICML 2025 · similarity 0.88)
- Multi-modal Topology-embedded Graph Learning for Spatially Resolved Genes Prediction from Pathology Images with Prior Gene Similarity Information (CVPR 2025 · similarity 0.88)
- SUICA: Learning Super-high Dimensional Sparse Implicit Neural Representations for Spatial Transcriptomics (ICML 2025 · similarity 0.88)
- STINR: Deciphering Spatial Transcriptomics via Implicit Neural Representation (CVPR 2025 · similarity 0.87)
- Diffusion Generative Modeling for Spatially Resolved Gene Expression Inference from Histology Images (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
