🧩 NRF2 reprogams BM progenitors into immunosuppressive macrophages
⚠️ High CIN fuels cancer non–cell-autonomously, promoting relapse
🗺️ STORIES learns "Waddington landscapes" from spatial transcriptomics
🔭 Early myeloid clusters predict if precancerous lesions regress/progress

⚡ CAFs act like neurons
➡️ velotest puts RNA velocity visualizations to the test
🧠 CytoTRACE 2 decodes absolute cellular potency
🧩 scooby predicts multiomic cell profiles from DNA
💉 mRNA vaccines boost immunotherapy response
🐋 Bowhead whales fight cancer with DNA repair

🧠 Large Perturbation Model predicts cellular responses in silico
🧩 MIMA unifies any omics with AI-driven batch correction
🧱 fastCNV maps tumor CNVs from single-cell & spatial data—fast ⚡ & accurate
🌸 PHLOWER untangles complex differentiation trees from multimodal data

🧓 Single-cell phylogenies map 70 years of clonal evolution in the human esophagus
💊 Generative AI designs new antibiotics effective against superbugs
🍄 Shorkie: a fungal DNA LM that predicts gene expression from sequence alone

This work introduces MIMA, a powerful AI framework for integrating spatial & single-cell multi-omics data while removing batch effects, revealing new molecular patterns & linking omics to histopathology.

Big step toward AI-driven digital pathology and precision oncology.

🐀 4 amino acids in mole-rat cGAS supercharge DNA repair
🔥 Tumor acidosis drives metabolic rewiring
🧠 Benchmarking 40 multimodal integration tools
📊 Self-supervised models face domain gaps in spatial data
🤖 scConcept, MrVI, & gReLU push single-cell AI & genomics design forward