Research map / Large language models / Retrieval and knowledge
ST-LLM: Spatial Transcriptomics Embedding with Large Language Models
Zhetao Xu, Xiaohua Wan, Le Li, Shuang Feng, Yiming Zhang, Fa Zhang, Bin Hu
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 |
|---|---|---|---|
| replace traditional GNN/VAE/diffusion embeddings with ST-LLM LLM embeddings | spatial transcriptomics clustering · DLPFC mean (10 slices) | ARI: 0.534 -> 0.572 | Improved |
| use ST-LLM instead of SpaceFlow | DLPFC 151670 · DLPFC 151670 | ARI: 0.473 -> 0.438 | Hurt |
| use ST-LLM instead of GraphST | region detection · Mouse Brain Posterior region detection | accuracy: 0.823 -> 0.814 | Hurt |
| use ST-LLM instead of GraphST | region detection · Human Lymph Node region detection | accuracy: 0.905 -> 0.973 | Improved |
| use structured ST-LLM prompt instead of RAG prompt (ablation) | ST-LLM · Human Breast Cancer (prompt comparison) | AMI: 0.485 -> 0.635 | Improved |
| use structured prompt instead of secondary-embedding prompt (ablation) | ST-LLM · Human Breast Cancer (prompt comparison) | AMI: 0.157 -> 0.635 | Improved |
| add spatial, task and bio components to graph-only prompt (ablation) | ST-LLM · Human Breast Cancer (embedding ablation) | AMI: 0.473 -> 0.626 | Improved |
| add graph component to spatial-only prompt (ablation) | ST-LLM · Human Breast Cancer (embedding ablation) | accuracy: 0.356 -> 0.846 | Improved |
| swap LLM backbone from DeepSeek to Smollm (ablation) | ST-LLM · Human Breast Cancer (LLM backbone) | AMI: 0.636 -> 0.520 | Hurt |
| swap LLM backbone from DeepSeek to Qwen (ablation) | ST-LLM · Human Breast Cancer (LLM backbone) | AMI: 0.636 -> 0.623 | No clear effect |
Closest papers by meaning
- SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics (ICML 2025 · similarity 0.89)
- 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 0.87)
- Diffusion Generative Modeling for Spatially Resolved Gene Expression Inference from Histology Images (ICLR 2025 · similarity 0.87)
- STINR: Deciphering Spatial Transcriptomics via Implicit Neural Representation (CVPR 2025 · similarity 0.86)
- Global Context-aware Representation Learning for Spatially Resolved Transcriptomics (ICML 2025 · similarity 0.86)
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.
