Deep contrastive learning enables genome-wide virtual screening DOI Creative Commons
Yinjun Jia, Bowen Gao, Jiaxin Tan

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Сен. 3, 2024

Abstract Numerous protein-coding genes are associated with human diseases, yet approximately 90% of them lack targeted therapeutic intervention. While conventional computational methods such as molecular docking have facilitated the discovery potential hit compounds, development genome-wide virtual screening against expansive chemical space remains a formidable challenge. Here we introduce DrugCLIP, novel framework that combines contrastive learning and dense retrieval to achieve rapid accurate screening. Compared traditional methods, DrugCLIP improves speed by several orders magnitude. In terms performance, not only surpasses other deep learning-based across two standard benchmark datasets but also demonstrates high efficacy in wet-lab experiments. Specifically, successfully identified agonists < 100 nM affinities for 5HT 2A R, key target psychiatric diseases. For another NET, whose structure is newly solved included training set, our method achieved rate 15%, 12 diverse molecules exhibiting better than Bupropion. Additionally, chemically inhibitors were validated determination Cryo-EM. Building on this foundation, present results pioneering trillion-scale screening, encompassing 10,000 AlphaFold2 predicted proteins within genome 500 million from ZINC Enamine REAL database. This work provides an innovative perspective drug post-AlphaFold era, where comprehensive targeting all disease-related reach.

Язык: Английский

AlphaFold and what is next: bridging functional, systems and structural biology DOI Creative Commons
Kacper Szczepski, Łukasz Jaremko

Expert Review of Proteomics, Год журнала: 2025, Номер unknown

Опубликована: Янв. 17, 2025

The DeepMind's AlphaFold (AF) has revolutionized biomedical research by providing both experts and non-experts with an invaluable tool for predicting protein structures. However, while AF is highly effective structures of rigid globular proteins, it not able to fully capture the dynamics, conformational variability, interactions proteins ligands other biomacromolecules. In this review, we present a comprehensive overview latest advancements in 3D model predictions biomacromolecules using AF. We also provide detailed analysis its strengths limitations, explore more recent iterations, modifications, practical applications strategy. Moreover, map path forward expanding landscape toward every peptide proteome most physiologically relevant form. This discussion based on extensive literature search performed PubMed Google Scholar. While significant progress been made enhance AF's modeling capabilities, argue that combined approach integrating various silico vitro methods will be beneficial future structural biology, bridging gaps between static dynamic features their functions.

Язык: Английский

Процитировано

0
Ligand-Conditioned Side Chain Packing for Flexible Molecular Docking DOI
Ding Luo, Xiaoyang Qu,

Dexin Lu

и другие.

Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Янв. 25, 2025

Molecular docking is a crucial technique for elucidating protein-ligand interactions. Machine learning-based methods offer promising advantages over traditional approaches, with significant potential further development. However, many current machine face challenges in ensuring the physical plausibility of generated poses. Additionally, accommodating protein flexibility remains difficult existing methods, limiting their effectiveness real-world scenarios. Herein, we present ApoDock, modular paradigm that combines learning-driven conditional side chain packing based on backbone and ligand information sampling to ensure physically realistic The poses are finally scored by developed mixture density network-based scoring function. With accurate packing, physical-based pose sampling, ranking ability, ApoDock demonstrates competitive performance across diverse applications, especially when using modeled structure (AlphaFold2 ESMFold) docking, exhibiting success rate 28.5% higher than other state art (SOTA), highlighting its as valuable tool binding studies related applications.

Язык: Английский

Процитировано

0
New strategies to enhance the efficiency and precision of drug discovery DOI Creative Commons

Qi An,

Liang Huang, Chuan Wang

и другие.

Frontiers in Pharmacology, Год журнала: 2025, Номер 16

Опубликована: Фев. 11, 2025

Drug discovery plays a crucial role in medicinal chemistry, serving as the cornerstone for developing new treatments to address wide range of diseases. This review emphasizes significance advanced strategies, such Click Chemistry, Targeted Protein Degradation (TPD), DNA-Encoded Libraries (DELs), and Computer-Aided Design (CADD), boosting drug process. Chemistry streamlines synthesis diverse compound libraries, facilitating efficient hit lead optimization. TPD harnesses natural degradation pathways target previously undruggable proteins, while DELs enable high-throughput screening millions compounds. CADD employs computational methods refine candidate selection reduce resource expenditure. To demonstrate utility these methodologies, we highlight exemplary small molecules discovered past decade, along with summary marketed drugs investigational that exemplify their clinical impact. These examples illustrate how techniques directly contribute advancing chemistry from bench bedside. Looking ahead, Artificial Intelligence (AI) technologies interdisciplinary collaboration are poised growing complexity discovery. By fostering deeper understanding transformative this aims inspire innovative research directions further advance field chemistry.

Язык: Английский

Процитировано

0
DeepPath: Overcoming data scarcity for protein transition pathway prediction using physics-based deep learning DOI Creative Commons
Yui Tik Pang,

Katie M. Kuo,

Lixinhao Yang

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Март 2, 2025

The structural dynamics of proteins play a crucial role in their function, yet most experimental and deep learning methods produce only static models. While molecular (MD) simulations provide atomistic insight into conformational transitions, they remain computationally prohibitive, particularly for large-scale motions. Here, we introduce DeepPath, deep-learning-based framework that rapidly generates physically realistic transition pathways between known protein states. Unlike conventional supervised approaches, DeepPath employs active to iteratively refine its predictions, leveraging mechanical force fields as an oracle guide pathway generation. We validated on three biologically relevant test cases: SHP2 activation, CdiB H1 secretion, the BAM complex lateral gate opening. accurately predicted all cases, reproducing key intermediate structures transient interactions observed previous studies. Notably, also inwardand outward-open states closely aligns with experimentally hybrid-barrel structure (TMscore = 0.91). Across achieved accurate predictions within hours, showcasing efficient alternative MD exploring transitions.

Язык: Английский

Процитировано

0
UNC9426, a Potent and Orally Bioavailable TYRO3-Specific Inhibitor DOI
Deyu Kong,

Xiangbo Yang,

Samantha Judd

и другие.

Journal of Medicinal Chemistry, Год журнала: 2025, Номер unknown

Опубликована: Март 11, 2025

TYRO3 plays a critical role in platelet aggregation as response amplifier. Selective inhibition of may provide therapeutic benefits for treating thrombosis and related diseases without increasing bleeding risk. We employed structure-based approach discovered novel potent inhibitor UNC9426 (12) with an excellent Ambit selectivity score (S50 (1.0 μM) = 0.026) favorable pharmacokinetic properties mice. Treatment reduced time blocked TYRO3-dependent functions tumor cells macrophages, implicating its utility multiple indications.

Язык: Английский

Процитировано

0
AI/ML methodologies and the future-will they be successful in designing the next generation of new chemical entities? DOI Creative Commons
Rachelle J. Bienstock

Journal of Cheminformatics, Год журнала: 2025, Номер 17(1)

Опубликована: Апрель 6, 2025

Cheminformatics and chemical databases are essential to drug discovery. However, machine learning (ML) artificial intelligence (AI) methodologies changing the way in which data is used. How will use of change discovery moving forward? do new ML methods molecular property prediction, hit lead target identification structure prediction differ compare with previous computational methods? Will improve diversity ligand design, offer enhancements. There still many advantages physics based they something lacking ML/ AI methods. Additionally, training often give best results when experimental assay measurements fed back into model. Often modeling not diametrically opposed but greatest advantage used complementary.

Язык: Английский

Процитировано

0
Investigations on genomic, topological and structural properties of diguanylate cyclases involved in Vibrio cholerae biofilm signalling using in silico techniques: Promising drug targets in combating cholera DOI Creative Commons

Tuhin Manna,

Subhamoy Dey,

Monalisha Karmakar

и другие.

Current Research in Structural Biology, Год журнала: 2025, Номер unknown, С. 100166 - 100166

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0
Deep contrastive learning enables genome-wide virtual screening DOI Creative Commons
Yinjun Jia, Bowen Gao, Jiaxin Tan

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Сен. 3, 2024

Abstract Numerous protein-coding genes are associated with human diseases, yet approximately 90% of them lack targeted therapeutic intervention. While conventional computational methods such as molecular docking have facilitated the discovery potential hit compounds, development genome-wide virtual screening against expansive chemical space remains a formidable challenge. Here we introduce DrugCLIP, novel framework that combines contrastive learning and dense retrieval to achieve rapid accurate screening. Compared traditional methods, DrugCLIP improves speed by several orders magnitude. In terms performance, not only surpasses other deep learning-based across two standard benchmark datasets but also demonstrates high efficacy in wet-lab experiments. Specifically, successfully identified agonists < 100 nM affinities for 5HT 2A R, key target psychiatric diseases. For another NET, whose structure is newly solved included training set, our method achieved rate 15%, 12 diverse molecules exhibiting better than Bupropion. Additionally, chemically inhibitors were validated determination Cryo-EM. Building on this foundation, present results pioneering trillion-scale screening, encompassing 10,000 AlphaFold2 predicted proteins within genome 500 million from ZINC Enamine REAL database. This work provides an innovative perspective drug post-AlphaFold era, where comprehensive targeting all disease-related reach.

Язык: Английский

Процитировано

0