Song Wu

Established a comprehensive research framework (Data Resource-Intelligent Tool-Mechanism Analysis) for systematic research on cancer alternative splicing, encompassing ASCancer Atlas, SpliceAgent and AS-CRC.

In this study, I compiled experimentally validated cancer-associated splicing events and systematically analyzed large-scale datasets from TCGA pan-cancer and GTEx pan-tissue cohorts to characterize splicing patterns. Additionally, I developed multi-dimensional online tools that integrate multi-omics and clinical data to identify splicing regulators, assess clinical relevance, and evaluate functional impacts through drug and protein interaction analyses. This resource equips researchers with comprehensive knowledge, data, and tools to understand cancer-specific splicing dysregulation.

In this study, I exploratively applied large language models (LLMs) and AI Agent technologies to cancer aberrant splicing research, developing the SpliceAgent. It features three core modules: 1) an intelligent curation assistant based on LLMs; 2) a cancer splicing Q&A chatbot integrating RAG, Graph-RAG, and real-time web retrieval; 3) an agentic analysis workflow automating data-to-report processes.

In this study, I gained extensive expertise in spatial transcriptomics through analyzing over 1,000 diverse samples across multiple spatial technologies. By creating a unified analytical platform for these datasets, I significantly enhanced research capabilities for investigating tissue structure and spatial gene expression patterns in life sciences.