- This event has passed.
AMS Special Seminar Series | Alan Amin
Location: Shaffer 3
When: January 22nd at 1:30 p.m.
Title: Machine Learning for Nature’s Overlooked Experiments
Abstract: Large-scale efforts have amassed vast biological sequence data, enabling machine learning models to learn conserved motifs and advance protein engineering. Yet in many high-impact domains—shrinking proteins for therapeutic delivery, engineering antibodies for diverse targets, incorporating priors into underpowered genetic studies—practitioners still resort to mining homologues, starting from scratch, or using simple linear models. The barrier is not data scarcity but that models trained on canonical evolutionary signals fail when the relevant information lies elsewhere. My work demonstrates that large-scale biological data can be leveraged for these problems given the appropriate statistical and computational framework: learning how nature navigates combinatorial deletion space to build smaller proteins, how our immune systems optimize antibodies, and how flexible models can scale to massive genetic datasets without overfitting.
Zoom link: https://wse.zoom.us/j/92755277282?pwd=iULpLaFnWAcWl6tQYUbeyZaN3zwBzn.1