Project write-ups and notes.
Designing CAR antigen-binding domains against the extracellular region of CD20 for the BioML Challenge 2024, using a pipeline of RFDiffusion backbone generation, ProteinMPNN inverse folding, AlphaFold2 filtering, and Chai-1 complex validation.
A Caves Lab-sponsored pipeline that uses high-resolution 2D photos from the NSF-funded Big Bee Project to estimate visual acuity across bee species by measuring ommatidia diameter and interommatidial angles. Initial results in the Apidae family show that male bees have larger ommatidia diameters while female bees have better CPD.
A UCSB Directed Reading Program project on whether neural networks can approximate reasonable functions to an arbitrary degree of accuracy. We studied the paper by Leshno et al. showing that, under modest assumptions, a broad class of activation functions are suitable for building neural networks to approximate continuous functions.