6-Parameter Model Maps Molecular Co-Folding to Cell Phenotypes
dom_beaini · x · 2026-08-26
The author built a minimal "toy" Virtual Cell model with only 6 trainable parameters. It combines 34 million protein-ligand and protein-protein co-foldings with gene expression and phenomics maps from Recursion. Despite its simplicity, the biologically grounded model achieved an average AUROC of 71% across 11 blinded cell lines. This prototype demonstrates how predicted molecular interactions can be mapped to cellular phenotypes.
More from Research
- LightSuite: fast cell detection and whole-brain atlas registration on a standard workstation — hisspikeness · 2026-10-15
- Pathway CSO pitches BDH as post-Transformer architecture uniting memory, reasoning, adaptation — schwentker · 2026-10-15
- AI-designed molecular scissors cut Alzheimer's proteins — but is it really AI? — gerardsans · 2026-10-15
- EmbodiedEvo tops RoboDojo at 38.95% success by letting robots inherit failed practice — thetripathi58 · 2026-10-15
- 797GB of n-gram counts in CPU RAM let 1B models match 57% more compute — PangWeiKoh · 2026-10-15
- Google's HARNESS² evolves agent harnesses without verifiers, up to +11.5 points on WorkBuddy-Bench — dair_ai · 2026-10-15