AMix-1: A Pathway to Test-Time Scalable Protein Foundation Model

TL;DR

AMix-1 is a scalable protein foundation model that leverages Bayesian Flow Networks, in-context learning, and test-time scaling to enhance protein design and engineering, achieving significant activity improvements and scalable performance.

Contribution

We present AMix-1, a novel protein foundation model with a systematic training methodology, in-context learning framework, and test-time scaling algorithm for scalable protein engineering.

Findings

Achieved a 1.7-billion parameter model with structural understanding.

Designed a protein variant with up to 50x activity increase.

Demonstrated scalable performance gains with test-time algorithms.

Abstract

We introduce AMix-1, a powerful protein foundation model built on Bayesian Flow Networks and empowered by a systematic training methodology, encompassing pretraining scaling laws, emergent capability analysis, in-context learning mechanism, and test-time scaling algorithm. To guarantee robust scalability, we establish a predictive scaling law and reveal the progressive emergence of structural understanding via loss perspective, culminating in a strong 1.7-billion model. Building on this foundation, we devise a multiple sequence alignment (MSA)-based in-context learning strategy to unify protein design into a general framework, where AMix-1 recognizes deep evolutionary signals among MSAs and consistently generates structurally and functionally coherent proteins. This framework enables the successful design of a dramatically improved AmeR variant with an up to $50\times$ activity increase over its wild type. Pushing the boundaries of protein engineering, we further empower AMix-1 with an evolutionary test-time scaling algorithm for in silico directed evolution that delivers substantial, scalable performance gains as verification budgets are intensified, laying the groundwork for next-generation lab-in-the-loop protein design.