An agnostic analysis of the human AlphaFold2 proteome using local protein conformations

Brevern, Alexandre G. de

doi:10.1016/j.biochi.2022.11.009

Cited by 7 publications

(9 citation statements)

References 100 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…175 Structural alphabets also appear as a means to analyze underrepresented secondary structure elements that could be badly scored in terms of pLDDT but correct in terms of structure. 176 As mentioned above for membrane proteins, alternative tools can be used for validations, such as DREAM or MembraneFold/DeepTMHMM, 177,178 to identify elements associated with the membrane, or MD to test the stability of the system. The latter could also, in principle, improve the model.…”

Section: ■ Modeling Validationmentioning

confidence: 99%

“…In addition to the possibility of analyzing models with a biochemist specialist of the protein, as mentioned above, it could be worth analyzing the basic properties of the model, such as clashes, Ramachandran plot outliers, etc. , This also opens the possibility of the model’s improvements via molecular dynamics simulations , or methods directly based on angles . Structural alphabets also appear as a means to analyze underrepresented secondary structure elements that could be badly scored in terms of pLDDT but correct in terms of structure . As mentioned above for membrane proteins, alternative tools can be used for validations, such as DREAM or MembraneFold/DeepTMHMM, , to identify elements associated with the membrane, or MD to test the stability of the system.…”

Section: Modeling Validationmentioning

confidence: 99%

See 1 more Smart Citation

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

Versini,

Sritharan,

Aykac Fas

et al. 2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

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.

show abstract

Section: ■ Modeling Validationmentioning

confidence: 99%

Section: Modeling Validationmentioning

confidence: 99%

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

Versini,

Sritharan,

Aykac Fas

et al. 2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

show abstract

“…The recent release of DeepMind’s AlphaFold2 (version 2.1.0) (AF2) [ 48 , 49 ] and AlphaFold-Multimer (version 2.1.0) (AFM2) [ 50 , 51 ] has brought the accuracy of the computational modeling of proteins to another level. Several studies have shown that both AFM2 and the input-manipulated versions of AF2 are able to predict protein–peptide complexes with high accuracy [ 51 , 52 , 53 , 54 ]. However, they have several major limitations including their (i) protein-only predictions, excluding cofactors, ions, or any post-translational modifications, (ii) inconsistency in the prediction quality of secondary structures and other local conformations due to their over- and under-representations during the training process [ 52 , 53 ], (iii) complete neglect of the effect of critical water molecules at the binding interface, (iv) decreased prediction accuracy for protein side-chains [ 54 ], and (v) inadequate modeling of conformational flexibility [ 49 , 55 ] which is crucial for modeling ligand binding with induced fit.…”

Section: Introductionmentioning

confidence: 99%

“…Several studies have shown that both AFM2 and the input-manipulated versions of AF2 are able to predict protein–peptide complexes with high accuracy [ 51 , 52 , 53 , 54 ]. However, they have several major limitations including their (i) protein-only predictions, excluding cofactors, ions, or any post-translational modifications, (ii) inconsistency in the prediction quality of secondary structures and other local conformations due to their over- and under-representations during the training process [ 52 , 53 ], (iii) complete neglect of the effect of critical water molecules at the binding interface, (iv) decreased prediction accuracy for protein side-chains [ 54 ], and (v) inadequate modeling of conformational flexibility [ 49 , 55 ] which is crucial for modeling ligand binding with induced fit. A recent study also showed that despite the excellent structural agreement of their predicted ligand bound conformation to the experimental one, deep learning-based docking methods often produce physically implausible structures [ 56 ] and can be outperformed by standard, physics-based docking methods.…”

Section: Introductionmentioning

confidence: 99%

Efficient Refinement of Complex Structures of Flexible Histone Peptides Using Post-Docking Molecular Dynamics Protocols

Bayarsaikhan,

Zsidó,

Börzsei

et al. 2024

IJMS

View full text Add to dashboard Cite

Histones are keys to many epigenetic events and their complexes have therapeutic and diagnostic importance. The determination of the structures of histone complexes is fundamental in the design of new drugs. Computational molecular docking is widely used for the prediction of target–ligand complexes. Large, linear peptides like the tail regions of histones are challenging ligands for docking due to their large conformational flexibility, extensive hydration, and weak interactions with the shallow binding pockets of their reader proteins. Thus, fast docking methods often fail to produce complex structures of such peptide ligands at a level appropriate for drug design. To address this challenge, and improve the structural quality of the docked complexes, post-docking refinement has been applied using various molecular dynamics (MD) approaches. However, a final consensus has not been reached on the desired MD refinement protocol. In this present study, MD refinement strategies were systematically explored on a set of problematic complexes of histone peptide ligands with relatively large errors in their docked geometries. Six protocols were compared that differ in their MD simulation parameters. In all cases, pre-MD hydration of the complex interface regions was applied to avoid the unwanted presence of empty cavities. The best-performing protocol achieved a median of 32% improvement over the docked structures in terms of the change in root mean squared deviations from the experimental references. The influence of structural factors and explicit hydration on the performance of post-docking MD refinements are also discussed to help with their implementation in future methods and applications.

show abstract

“…The prediction of protein structure has been an on-going challenge for computational methods, including artificial intelligence. However, the recent development of structure prediction deep learning (DL) tools such as alphafold2 (Jumper et al, 2021), ESMfold (Lin et al, 2022) or ProteinMPNN (Dauparas et al, 2022), has the potential to revolutionize this area (de Brevern, 2022;Goulet and Cambillau, 2022). Nevertheless, these DL tools are not suitable for predicting how individual amino acid changes alter protein structure and function (Eisenstein, 2021): they can't predict epistatic effects.…”

mentioning

confidence: 99%

Editorial: Machine learning, epistasis, and protein engineering: From sequence-structure-function relationships to regulation of metabolic pathways

Cadet

Saavedra

Syrén

et al. 2022

Front. Mol. Biosci.

View full text Add to dashboard Cite

Editorial on the Research Topic Machine learning, epistasis, and protein engineering: From sequencestructure-function relationships to regulation of metabolic pathways Epistasis is a term originating from genomics and describes the non-additivity of effects of gene interactions on functional parameters (Phillips, 2008). Both within and between genes, epistasis plays a fundamental role in the ability of protein sequences to evolve (Breen et al., 2012).In protein sciences, epistasis reflects non-linearity effects, ie., non-additive impacts, resulting from interactions between mutations within a protein sequence (Reetz, 2013). The evolution of proteins cannot be understood without the knowledge of the epistasis phenomena (non-additive mutational effects) that take place within them. Indeed, epistasis can reverse the effect of a mutation from beneficial to deleterious. Likewise, thanks to epistasis phenomena, the conservation of a neutral mutation during evolution can lead to beneficial effects of greater amplitude. Interactions between mutated amino acids, as well as intramolecular interaction networks that can be set up following mutations, condition the function (Acevedo-Rocha et al., 2021). While it is clear that understanding sequence-structure-function relationships, and in particular

show abstract

An agnostic analysis of the human AlphaFold2 proteome using local protein conformations

Cited by 7 publications

References 100 publications

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

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

Efficient Refinement of Complex Structures of Flexible Histone Peptides Using Post-Docking Molecular Dynamics Protocols

Editorial: Machine learning, epistasis, and protein engineering: From sequence-structure-function relationships to regulation of metabolic pathways

Contact Info

Product

Resources

About