Advancing structural biology through breakthroughs in AI

Aithani, Laksh; Alcaide, Eric; Bartunov, Sergey; Cooper, C.D.O.; Dore, A.S.; Lane, Thomas J; Maclean, Finlay; Rucktooa, Prakash; Shaw, Robert A.; Skerratt, Sarah E.

doi:10.1016/j.sbi.2023.102601

Cited by 13 publications

(4 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The level of conformational variability of different proteins can range from relatively rigid molecules to fully unstructured ones [ 1 ]. For more rigid proteins, a single three-dimensional (3D) structure, obtainable by experimental [ 2 ] or computational prediction methods [ 3 , 4 ], often provides key insights into its function. On the other hand, for highly dynamic proteins, function may only be fully understood by knowing the statistical properties of their structural ensembles [ 5 , 6 ].…”

Section: Introductionmentioning

confidence: 99%

Transferable deep generative modeling of intrinsically disordered protein conformations

Janson,

Feig

2024

PLoS Comput Biol

View full text Add to dashboard Cite

Intrinsically disordered proteins have dynamic structures through which they play key biological roles. The elucidation of their conformational ensembles is a challenging problem requiring an integrated use of computational and experimental methods. Molecular simulations are a valuable computational strategy for constructing structural ensembles of disordered proteins but are highly resource-intensive. Recently, machine learning approaches based on deep generative models that learn from simulation data have emerged as an efficient alternative for generating structural ensembles. However, such methods currently suffer from limited transferability when modeling sequences and conformations absent in the training data. Here, we develop a novel generative model that achieves high levels of transferability for intrinsically disordered protein ensembles. The approach, named idpSAM, is a latent diffusion model based on transformer neural networks. It combines an autoencoder to learn a representation of protein geometry and a diffusion model to sample novel conformations in the encoded space. IdpSAM was trained on a large dataset of simulations of disordered protein regions performed with the ABSINTH implicit solvent model. Thanks to the expressiveness of its neural networks and its training stability, idpSAM faithfully captures 3D structural ensembles of test sequences with no similarity in the training set. Our study also demonstrates the potential for generating full conformational ensembles from datasets with limited sampling and underscores the importance of training set size for generalization. We believe that idpSAM represents a significant progress in transferable protein ensemble modeling through machine learning.

show abstract

Section: Introductionmentioning

confidence: 99%

Transferable deep generative modeling of intrinsically disordered protein conformations

Janson,

Feig

2024

PLoS Comput Biol

View full text Add to dashboard Cite

show abstract

“…Second-generation genome sequencing can rapidly provide a large number of complex and highly repetitive genomic sequences and relevant annotation information [4] . With the emergence of deep learning-based structural prediction tools, such as AlphaFold2 and RoseTTAFold, a massive number of sequences have been transformed into billions of protein structures, further elucidating the relationship between structure and function [5] . Therefore, the study of the metabolic potential of biocontrol fungi based on sequence and structural omics data has become an increasingly interesting and important topic [6] .…”

Section: Introductionmentioning

confidence: 99%

Unraveling the metabolic potential of biocontrol fungi through omics data: a key to enhancing large-scale application strategies

Yang,

Wu,

Sun

et al. 2024

ABBS

View full text Add to dashboard Cite

Biological control of pests and pathogens has attracted much attention due to its green, safe and effective characteristics. However, it faces the dilemma of insignificant effects in large-scale applications. Therefore, an in-depth exploration of the metabolic potential of biocontrol fungi based on big omics data is crucial for a comprehensive and systematic understanding of the specific modes of action operated by various biocontrol fungi. This article analyzes the preferences for extracellular carbon and nitrogen source degradation, secondary metabolites (nonribosomal peptides, polyketide synthases) and their product characteristics and the conversion relationship between extracellular primary metabolism and intracellular secondary metabolism for eight different filamentous fungi with characteristics appropriate for the biological control of bacterial pathogens and phytopathogenic nematodes. Further clarification is provided that Paecilomyces lilacinus , encoding a large number of hydrolase enzymes capable of degrading pathogen protection barrier, can be directly applied in the field as a predatory biocontrol fungus, whereas Trichoderma , as an antibiosis-active biocontrol control fungus, can form dominant strains on preferred substrates and produce a large number of secondary metabolites to achieve antibacterial effects. By clarifying the levels of biological control achievable by different biocontrol fungi, we provide a theoretical foundation for their application to cropping habitats.

show abstract

“…32 OmegaFold comes close to the performance of AlphaFold2 in general, sometimes surpassing it on orphan sequences, by using a language model trained on amino acid sequences instead of the direct use of MSA. 33 Predicting mutation-induced stability changes is crucial for protein design and precision medicine. In contrast, there is a serious lack of protein 3D structure, which leads to poor prediction model accuracy and generalization ability.…”

Section: ■ Introductionmentioning

confidence: 99%

“…OmegaFold performs true ab initio folding without the use of MSA, the advantages of which include (1) that it does not rely on known protein structures, which means that it is able to predict protein structures without any prior structural knowledge; and (2) that it has the possibility of discovering new types of protein structures . OmegaFold comes close to the performance of AlphaFold2 in general, sometimes surpassing it on orphan sequences, by using a language model trained on amino acid sequences instead of the direct use of MSA …”

Section: Introductionmentioning

confidence: 99%

OmeDDG: Improved Protein Mutation Stability Prediction Based on Predicted 3D Structures

Liu,

Jiang,

Yang

et al. 2023

J. Phys. Chem. B

View full text Add to dashboard Cite

Determining changes in the protein's thermal stability following mutations is critical in protein engineering and understanding pathogenic missense mutations. Despite the development of various computational methods to predict the effects of single-point mutations, their accuracy remains limited. In this study, we propose a new computational method, OmeDDG, that more accurately predicts mutation-induced Gibbs free energy changes in protein folding (ΔΔG). OmeDDG takes the sequences of wild-type and mutant proteins as input, utilizes OmegaFold to obtain the 3D structure, employs a convolutional neural network to extract structural features, and combines them with protein mutation features and pretraining features to predict the stability of single-point mutations in proteins. We performed a comprehensive comparison between OmeDDG and other available prediction methods on four blind test datasets, confirming that OmeDDG can effectively enhance protein mutation prediction performance. Notably, on the antisymmetric dataset Ssym, OmeDDG achieves the best performance, demonstrating favorable antisymmetry with PCC = 0.79 and RMSE = 0.96 for forward mutations and PCC = 0.77 and RMSE = 0.97 for reverse mutant types.

show abstract

Advancing structural biology through breakthroughs in AI

Cited by 13 publications

References 42 publications

Transferable deep generative modeling of intrinsically disordered protein conformations

Transferable deep generative modeling of intrinsically disordered protein conformations

Unraveling the metabolic potential of biocontrol fungi through omics data: a key to enhancing large-scale application strategies

OmeDDG: Improved Protein Mutation Stability Prediction Based on Predicted 3D Structures

Contact Info

Product

Resources

About