MAP Framework: Knowledge-Guided Single-Cell AI Predicts Drug Responses
bravo_abad · x · 2026-08-27
Jinghao Feng et al. introduce MAP, a single-cell AI framework for zero-shot drug response prediction. Instead of relying solely on measured data, MAP injects structured biological knowledge. The team built MAP-KG, integrating 14 public resources linking 187k drugs and 22k genes. Knowledge-driven pretraining aligns molecular structures, protein sequences, and text descriptions to predict transcriptional responses.
More from Research
- Hypothesis: Models May Learn to Conceal CoT to Evade Auditing — DimitrisPapail · 2026-08-27
- Nature Paper Proposes 4D 'Agentic Profiles' for AI Governance — Dr_Atoosa · 2026-08-27
- Recommended Quanta oral history: How ChatGPT reshaped the NLP field — voooooogel · 2026-08-27
- Anthropic study: 72% of AI conversations remain human-led — Diyi_Yang · 2026-08-27
- OpenAI swarm developed ethics: attacking infrastructure OK, humans not — morqon · 2026-08-27
- New Research Defines Criteria for Physical Computation and Neuromorphic Reservoirs — neurovium · 2026-08-27