DynaVelo uses neural ODEs to predict gene perturbations that rescue cell mutations
bravo_abad · x · 2026-10-22
Karbalayghareh and colleagues introduce DynaVelo, a model tackling the core limitation of single-cell biology: measurements destroy cells, yielding snapshots of different cells rather than a recording of the same cell developing.
- Method: combines gene expression (scRNA-seq) with DNA accessibility (scATAC-seq) via a neural ordinary differential equation that learns how cell states change over time, guided by RNA velocity estimates.
- Key capability: researchers can alter a gene's activity inside the trained model and observe how predicted dynamics shift, enabling in silico searches for second-site changes that compensate for a mutation.
- Application: if a mutation disrupts immune cell development (e.g., GC B cells), the model can predict which gene change might restore normal behavior, guiding experiments.
More from Research
- EMNLP 2026 industry track: keynotes on agentic challenges and constitutional alignment — verena_rieser · 2026-10-22
- Veteran reviewer: 'safe' cockroach papers get accepted, reviewers should reward flawed insight — plopesresearch · 2026-10-22
- Hyperparameter Tuning Is Itself a Machine-Learning Problem: Grid vs Random vs Bayesian — bravo_abad · 2026-10-22
- Michael Levin polls the public: how do physical, living, and minds overlap? — drmichaellevin · 2026-10-22
- ICLR reviewer: 2 of 3 assigned papers are complete AI slop, worst batch ever seen — AkariAsai · 2026-10-22
- Training-free 20B model Viveki generates real-time video heatmaps under 150ms — not-ekalabya · 2026-10-22