TRILL: ORCHESTRATING MODULAR DEEP-LEARNING WORKFLOWS FOR DEMOCRATIZED, SCALABLE PROTEIN ANALYSIS AND ENGINEERING

Opportunities and Challenges for Machine Learning-Assisted Enzyme Engineering

Enzymes can be engineered at the level of their amino acid sequences to optimize key properties such as expression, stability, substrate range, and catalytic efficiency-or even to unlock new catalytic activities not found in nature. Because the search space of possible proteins is vast, enzyme engineering usually involves discovering an enzyme starting point that has some level of the desired activity followed by directed evolution to improve its "fitness" for a desired application. Recently, machine learning (ML) has emerged as a powerful tool to complement this empirical process. ML models can contribute to (1) starting point discovery by functional annotation of known protein sequences or generating novel protein sequences with desired functions and (2) navigating protein fitness landscapes for fitness optimization by learning mappings between protein sequences and their associated fitness values. In this Outlook, we explain how ML complements enzyme engineering and discuss its future potential to unlock improved engineering outcomes.

A Perspective on the Prospective Use of AI in Protein Structure Prediction

AlphaFold2 (AF2) and RoseTTaFold (RF) have revolutionized structural biology, serving as highly reliable and effective methods for predicting protein structures. This article explores their impact and limitations, focusing on their integration into experimental pipelines and their application in diverse protein classes, including membrane proteins, intrinsically disordered proteins (IDPs), and oligomers. In experimental pipelines, AF2 models help X-ray crystallography in resolving the phase problem, while complementarity with mass spectrometry and NMR data enhances structure determination and protein flexibility prediction. Predicting the structure of membrane proteins remains challenging for both AF2 and RF due to difficulties in capturing conformational ensembles and interactions with the membrane. Improvements in incorporating membrane-specific features and predicting the structural effect of mutations are crucial. For intrinsically disordered proteins, AF2's confidence score (pLDDT) serves as a competitive disorder predictor, but integrative approaches including molecular dynamics (MD) simulations or hydrophobic cluster analyses are advocated for accurate dynamics representation. AF2 and RF show promising results for oligomeric models, outperforming traditional docking methods, with AlphaFold-Multimer showing improved performance. However, some caveats remain in particular for membrane proteins. Real-life examples demonstrate AF2's predictive capabilities in unknown protein structures, but models should be evaluated for their agreement with experimental data. Furthermore, AF2 models can be used complementarily with MD simulations. In this Perspective, we propose a "wish list" for improving deep-learning-based protein folding prediction models, including using experimental data as constraints and modifying models with binding partners or post-translational modifications. Additionally, a meta-tool for ranking and suggesting composite models is suggested, driving future advancements in this rapidly evolving field.

Unexplored regions of the protein sequence-structure map revealed at scale by a library of foldtuned language models

Nature has likely sampled only a fraction of all protein sequences and structures allowed by the laws of biophysics. However, the combinatorial scale of amino-acid sequence-space has traditionally precluded substantive study of the full protein sequence-structure map. In particular, it remains unknown how much of the vast uncharted landscape of far-from-natural sequences consists of alternate ways to encode the familiar ensemble of natural folds; proteins in this category also represent an opportunity to diversify candidates for downstream applications. Here, we characterize sequence-structure mapping in far-from-natural regions of sequence-space guided by the capacity of protein language models (pLMs) to explore sequences outside their natural training data through generation. We demonstrate that pretrained generative pLMs sample a limited structural snapshot of the natural protein universe, including >350 common (sub)domain elements. Incorporating pLM, structure prediction, and structure-based search techniques, we surpass this limitation by developing a novel "foldtuning" strategy that pushes a pretrained pLM into a generative regime that maintains structural similarity to a target protein fold (e.g. TIM barrel, thioredoxin, etc) while maximizing dissimilarity to natural amino-acid sequences. We apply "foldtuning" to build a library of pLMs for >700 naturally-abundant folds in the SCOP database, accessing swaths of proteins that take familiar structures yet lie far from known sequences, spanning targets that include enzymes, immune ligands, and signaling proteins. By revealing protein sequence-structure information at scale outside of the context of evolution, we anticipate that this work will enable future systematic searches for wholly novel folds and facilitate more immediate protein design goals in catalysis and medicine.