General Tabular Model Matches Or Beats Biology Foundation Models For Cell Perturbation Prediction
Commentary on X
Highest ranked
Also I fear there is something terribly wrong with the perturbation prediction field, with expt. designs (no notion of time, arbitrary end points) & performance metrics, if this thing can put all the models trained on massive single cell atlases & perturb datasets to bed. 2/2
11/ Honestly, I find this shocking and inspiring, as it shows what the future could hold...
Amazing work by @g_palla1 with support from Alex Hillsley and @yangjoonkim!
What is this super cool abomination, bat-from-hell perturbation prediction model 😁 (like in a good crazy way... what the actual f?!?). I truly hope this puts the "virtual cell" terminology for perturbation prediction models to rest in peace though. 😂😂😂 1/ https://twitter.com/loicaroyer/status/2074521242011635875