Exploring evolution to uncover insights into protein mutational stability DOI Creative Commons

Pauline Hermans,

Matsvei Tsishyn, Martin Schwersensky

et al.

Molecular Biology and Evolution, Journal Year: 2024, Volume and Issue: 42(1)

Published: Dec. 27, 2024

Determining the impact of mutations on thermodynamic stability proteins is essential for a wide range applications such as rational protein design and genetic variant interpretation.Since major driver evolution, evolutionary data are often used to guide predictions. Many state-of-the-art predictors extract information from multiple sequence alignments (MSA) homologous query protein, leverage it predict effects stability. To evaluate power limitations methods, we massive amount recently obtained by deep mutational scanning study how best construct MSAs optimally them. We tested different models found that, unexpectedly, independent-site achieve similar accuracy more complex epistatic models. A detailed analysis latter suggests that their inference results in noisy couplings, which do not appear add predictive over contribution, at least context prediction. Interestingly, combining any features with simple structural feature, relative solvent accessibility mutated residue, achieved prediction supervised, machine learning-based, change predictors. Our provide new insights into relationship between evolution stability, show can be exploited improve performance

Language: Английский

Revolutionizing Molecular Design for Innovative Therapeutic Applications through Artificial Intelligence DOI Creative Commons

Ahrum Son,

Jongham Park, Woojin Kim

et al.

Molecules, Journal Year: 2024, Volume and Issue: 29(19), P. 4626 - 4626

Published: Sept. 29, 2024

The field of computational protein engineering has been transformed by recent advancements in machine learning, artificial intelligence, and molecular modeling, enabling the design proteins with unprecedented precision functionality. Computational methods now play a crucial role enhancing stability, activity, specificity for diverse applications biotechnology medicine. Techniques such as deep reinforcement transfer learning have dramatically improved structure prediction, optimization binding affinities, enzyme design. These innovations streamlined process allowing rapid generation targeted libraries, reducing experimental sampling, rational tailored properties. Furthermore, integration approaches high-throughput techniques facilitated development multifunctional novel therapeutics. However, challenges remain bridging gap between predictions validation addressing ethical concerns related to AI-driven This review provides comprehensive overview current state future directions engineering, emphasizing their transformative potential creating next-generation biologics advancing synthetic biology.

Language: Английский

Citations

6
Assessing computational tools for predicting protein stability changes upon missense mutations using a new dataset DOI

Feifan Zheng,

Yang Liu, Yan Yang

et al.

Protein Science, Journal Year: 2023, Volume and Issue: 33(1)

Published: Dec. 12, 2023

Insight into how mutations affect protein stability is crucial for engineering, understanding genetic diseases, and exploring evolution. Numerous computational methods have been developed to predict the impact of amino acid substitutions on stability. Nevertheless, comparing these poses challenges due variations in their training data. Moreover, it observed that they tend perform better at predicting destabilizing than stabilizing ones. Here, we meticulously compiled a new dataset from three recently published databases: ThermoMutDB, FireProtDB, ProThermDB. This dataset, which does not overlap with well-established S2648 consists 4038 single-point mutations, including over 1000 mutations. We assessed using 27 methods, latest ones utilizing mega-scale datasets transfer learning. excluded entries or similarity ensure fairness. Pearson correlation coefficients tested tools ranged 0.20 0.53 unseen data, none could accurately even those performing well anti-symmetric property analysis. While most present consistent trends across various properties such as solvent exposure secondary conformation, do exhibit clear pattern. Our study also suggests solely addressing bias may significantly enhance accuracy These findings emphasize importance developing precise predictive

Language: Английский

Citations

11
Leveraging computer-aided design and artificial intelligence to develop a next-generation multi-epitope tuberculosis vaccine candidate DOI Creative Commons
Zhuang Li, Awais Ali,

Ling Yang

et al.

Infectious Medicine, Journal Year: 2024, Volume and Issue: 3(4), P. 100148 - 100148

Published: Nov. 9, 2024

Tuberculosis (TB) remains a global public health challenge. The existing Bacillus Calmette-Guérin vaccine has limited efficacy in preventing adult pulmonary TB, necessitating the development of new vaccines with improved protective effects.

Language: Английский

Citations

4
Shared-weight graph framework for comprehensive protein stability prediction across diverse mutation types DOI Creative Commons
Gen Li,

Sijie Yao,

Long Fan

et al.

Briefings in Bioinformatics, Journal Year: 2025, Volume and Issue: 26(2)

Published: March 1, 2025

Abstract Research on protein stability changes is vital for understanding disease mechanisms and optimizing industrial enzymes. Protein thermal can be modified by variants leading to in ΔΔG values between wild-type mutant proteins. Despite advances, most models focus single-point mutations, overlooking multipoint indel mutations. Typically, the mutation expected have a relatively limited impact function of protein, necessitating more drastic modifications meet new challenges. Current methods mutations yield poor results, no method exists any length To address this, we introduce UniMutStab, shared-graph convolutional network leveraging language residue interaction networks access type mutation. An embedded edge weight module enhances integration node features interactions, improving prediction accuracy. Trained “Mega-scale” dataset with ~780 000 UniMutStab surpasses existing predicting changes. It purely sequence-based approach predict arbitrary types, demonstrating robust generalization across multiple tasks potentially contributing significantly engineering, personalized therapeutics, diagnostic methodologies.

Language: Английский

Citations

0
Transfer Learning in Cancer Genetics, Mutation Detection, Gene Expression Analysis, and Syndrome Recognition DOI Open Access
Hamidreza Ashayeri, Navid Sobhi, Paweł Pławiak

et al.

Cancers, Journal Year: 2024, Volume and Issue: 16(11), P. 2138 - 2138

Published: June 4, 2024

Artificial intelligence (AI), encompassing machine learning (ML) and deep (DL), has revolutionized medical research, facilitating advancements in drug discovery cancer diagnosis. ML identifies patterns data, while DL employs neural networks for intricate processing. Predictive modeling challenges, such as data labeling, are addressed by transfer (TL), leveraging pre-existing models faster training. TL shows potential genetic improving tasks like gene expression analysis, mutation detection, syndrome recognition, genotype–phenotype association. This review explores the role of overcoming challenges expression, or phenotype–genotype shown effectiveness various aspects research. enhances accuracy efficiency aiding identification abnormalities. can improve diagnostic syndrome-related patterns. Moreover, plays a crucial analysis order to accurately predict levels their interactions. Additionally, association studies pre-trained models. In conclusion, AI prediction, detection. Future should focus on increasing domain similarities, expanding databases, incorporating clinical better predictions.

Language: Английский

Citations

3
The origin of mutational epistasis DOI
Jorge A. Vila

European Biophysics Journal, Journal Year: 2024, Volume and Issue: 53(7-8), P. 473 - 480

Published: Oct. 23, 2024

Language: Английский

Citations

1
Exploring evolution to enhance mutational stability prediction DOI Creative Commons

Pauline Hermans,

Matsvei Tsishyn, Martin Schwersensky

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: May 31, 2024

Abstract Determining the impact of mutations on thermodynamic stability proteins is essential for a wide range applications such as rational protein design and genetic variant interpretation. Since major driver evolution, evolutionary data are often used to guide predictions. Many state-of-the-art predictors extract information from multiple sequence alignments (MSA) homologous query protein, leverage it predict effects stability. To evaluate power limitations methods, we massive amount recently obtained by deep mutational scanning study how best construct MSAs optimally them. We tested different models found that, unexpectedly, independent-site achieve similar accuracy more complex epistatic models. A detailed analysis latter suggests that their inference results in noisy couplings, which do not appear add predictive over contribution, at least context prediction. Interestingly, combining any features with simple structural feature, relative solvent accessibility mutated residue, achieved prediction supervised, machine learning-based, change predictors. Our provide new insights into relationship between evolution stability, show can be exploited improve performance

Language: Английский

Citations

0
Enhancing protein stability prediction with geometric learning and pre-training strategies DOI
Minghui Li

Nature Computational Science, Journal Year: 2024, Volume and Issue: 4(11), P. 807 - 808

Published: Nov. 8, 2024

Language: Английский

Citations

0
Computational Protein Engineering DOI

Tuba Okur,

Onur Serçinoğlu

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

Language: Английский

Citations

0
Advances in Zero‐Shot Prediction‐Guided Enzyme Engineering Using Machine Learning DOI Open Access
Chang Liu, Junxian Wu, Yongbo Chen

et al.

ChemCatChem, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 23, 2024

Abstract The advent of machine learning (ML) has significantly advanced enzyme engineering, particularly through zero‐shot (ZS) predictors that forecast the effects amino acid mutations on properties without requiring additional labeled data for target enzyme. This review comprehensively summarizes ZS developed over past decade, categorizing them into kinetic parameters, stability, solubility/aggregation, and fitness. It details algorithms used, encompassing traditional ML approaches deep models, emphasizing their predictive performance. Practical applications in engineering specific enzymes are discussed. Despite notable advancements, challenges persist, including limited training necessity to incorporate environmental factors (e.g., pH, temperature) dynamics these models. Future directions proposed advance prediction‐guided thereby enhancing practical utility predictors.

Language: Английский

Citations

0