Abstract: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.

AI-Predicted Protein Deformation Encodes Energy Landscape Perturbation

AI-predicted protein deformation encodes energy landscape perturbation

Predicting protein conformational motions using energetic frustration analysis and AlphaFold2

State-of-the-Art Estimation of Protein Model Accuracy Using AlphaFold

AlphaFold2 can predict single-mutation effects

Predicting relative populations of protein conformations without a physics engine using AlphaFold2

AlphaFold2 knows some protein folding principles

Using AlphaFold to predict the impact of single mutations on protein stability and function

An Efficient Method to Predict Protein Thermostability in Alanine Mutation

Prediction of protein structure and AI

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

High-accuracy protein structures by combining machine-learning with physics-based refinement

The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins

AI told you so: navigating protein structure prediction in the era of machine learning

Protein structure prediction in the era of AI: Challenges and limitations when applying to in silico force spectroscopy

Effective High-Accuracy Prediction of Protein Structures from Easily Obtainable Artificial Homologous Sequences by Structure-Stability-Based Selection

AI-Based Protein Structure Prediction in Drug Discovery: Impacts and Challenges

Predicting Protein Flexibility with AlphaFold

Severe deviation in protein fold prediction by advanced AI: a case study