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.

Structures prediction and replica exchange molecular dynamics simulations of α-synuclein: A case study for intrinsically disordered proteins

A Coarse-Grained Molecular Dynamics Approach to the Study of the Intrinsically Disordered Protein α-Synuclein

In Silico Structural Analysis Exploring Conformational Folding of Protein Variants in Alzheimer's Disease

What Does Evolution Tell Us About The Structure Of A Functional Amyloid Protein?

Inherent Relationships among Different Biophysical Prediction Methods for Intrinsically Disordered Proteins

AlphaFold and Implications for Intrinsically Disordered Proteins

Assessing structure and disorder prediction tools for de novo emerged proteins in the age of machine learning

Navigating the Unstructured by Evaluating AlphaFold's Efficacy in Predicting Missing Residues and Structural Disorder in Proteins

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

A Unified De Novo Approach for Predicting the Structures of Ordered and Disordered Proteins

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

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

Conformational ensembles of the human intrinsically disordered proteome

Assessment of AI-Based Protein Structure Prediction for the NLRP3 Target

Exploring Intra- and Inter-Regional Interactions in the IDP α-Synuclein Using smFRET and MD Simulations

AlphaFold2 as a replacement for solution NMR structure determination of small proteins: Not so fast!

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

Protein structure prediction: Combining de novo modeling with sparse experimental data

A Case Study of Disulfide-Bond Impact on the Intrinsically Disordered Amyloid Beta

Tutorial: a guide for the selection of fast and accurate computational tools for the prediction of intrinsic disorder in proteins

Intrinsic Protein Disorder and Conditional Folding in AlphaFoldDB