The Proteome as a Lectome: Predictions from Deep Learning Propose Substantial Protein-Carbohydrate Interplay

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: March 12, 2025

It is a grand challenge to identify all the protein - carbohydrate interactions in an organism. Direct experiments would require extensive libraries of glycans definitively distinguish binding from non-binding proteins. Computational screening proteins for carbohydrate-binding provides attractive and ultimately testable alternative. Recent computational techniques have focused primarily on which residues interact with carbohydrates or species binds to. Current estimates label 1.5 5% as proteins; however, 50-70% are known be glycosylated, suggesting potential wealth that bind carbohydrates. We therefore developed novel dataset neural network architecture, named P rotein i nteraction Ca rbohydrates redictor (PiCAP), predict whether non-covalently carbohydrate. trained PiCAP binders, we selected identified likely not carbohydrates, including DNA-binding transcription factors, cytoskeletal components, antibodies, small-molecule-binding achieves 90% balanced accuracy protein-level predictions binding/non-binding. Using same dataset, model rbohydrate S ite I denti f ier 2 (CAPSIF2) CAPSIF2 Dice coefficient 0.57 residue-level our independent test outcompeting previous models this task. To demonstrate biological applicability CAPSIF2, investigated cell surface human cells further predicted likelihood three proteomes, notably E. coli, M. musculus , H. sapiens predicts approximately 35-40% these proteomes indicating substantial interplay protein-carbohydrate cellular functionality. The totality carbohydrate-protein remains elusive, part due inability versus glycomes high throughput manner. Here show first high-throughput methodology at proteomic scales by using structural sequence information. This information will allow scientists target better determine how elusive biomolecules play roles functions.

Language: Английский

---

A structural perspective on enzymes and their catalytic mechanisms

Current Opinion in Structural Biology, Journal Year: 2025, Volume and Issue: 92, P. 103040 - 103040

Published: March 31, 2025

In this perspective, we analyse the progress made in our knowledge of enzyme sequences, structures and functions last 2 years. We review how much new data have been garnered annotated, derived from study proteins using structural computational approaches. Recent advances towards capturing 'Catalysis silico' are described, including predictions structures, their interactions mechanisms. highlight flood data, driven by metagenomic sequencing, improved resources, high coverage Protein Data Bank E.C. classes AI-driven structure prediction techniques that facilitate accurate protein structures. note focus on disordered regions context regulation specificity comment emerging bioinformatic approaches capture reaction mechanisms computationally for comparing predicting also consider drivers field next five

Language: Английский

---

Quantum Mechanics Paradox in Protein Structure Prediction: Intrinsically Linked to Sequence yet Independent of it

Deleted Journal, Journal Year: 2025, Volume and Issue: unknown, P. 100039 - 100039

Published: April 1, 2025

Language: Английский

---

A Reversible Spherical Geometric Conversion of Protein Backbone Structure Coordinate Matrix to Three Independent Vectors of ρ, θ and φ

Published: April 8, 2024

Due to the vast conformational space proteins can adopt, accurate and efficient prediction of protein structure remains still a challenging task, coupled with intricacies interatomic interactions limitations current computational models in effectively navigating this complex molecular landscape. Additionally, lack comprehensive experimental data for all structures further exacerbates difficulty reliable machine learning-based three-dimensional conformations proteios building block life. Geometrically, Cartesian coordinate system (CCS, X, Y Z) spherical (SCS, ρ, θ φ) are two interconvertible systems, like sides one coin. Since beginning Protein Data Bank (PDB) 1971, CCS has been default approach specify atomic positions Z PDB. In manuscript, therefore, I present novel method reversible geometric conversion backbone matrices three independent vectors: φ. This facilitates lossless extraction essential structural features from data, enabling development advanced algorithms future. short, inter-atomic SCS offers yet means representing geometry, leveraging coordinates capture spatial relationships compact intuitive manner, provide robust framework feature ongoing efforts advancing field prediction, holy grail biology.

Language: Английский

---

Enhancing the synthesis efficiency of galacto-oligosaccharides of a β-galactosidase from Paenibacillus barengoltzii by engineering the active and distal sites

Food Chemistry, Journal Year: 2025, Volume and Issue: unknown, P. 144208 - 144208

Published: April 1, 2025

Language: Английский

---

Experimental and computational approaches for membrane protein insertion and topology determination

Methods, Journal Year: 2024, Volume and Issue: 226, P. 102 - 119

Published: April 9, 2024

Membrane proteins play pivotal roles in a wide array of cellular processes and constitute approximately quarter the protein-coding genes across all organisms. Despite their ubiquity biological significance, our understanding these remains notably less comprehensive compared to soluble counterparts. This disparity knowledge can be attributed, part, inherent challenges associated with employing specialized techniques for investigation membrane protein insertion topology. review will center on discussion molecular biology methodologies computational prediction tools designed elucidate topology helical proteins.

Language: Английский

---

Allostery Illuminated: Harnessing AI and Machine Learning for Drug Discovery

ACS Medicinal Chemistry Letters, Journal Year: 2024, Volume and Issue: 15(9), P. 1449 - 1455

Published: Aug. 30, 2024

In the past several years there has been rapid adoption of artificial intelligence (AI) and machine learning (ML) tools for drug discovery. this Microperspective, we comment on recent AI/ML applications to discovery allosteric modulators, focusing breakthroughs with AlphaFold, structure-based (SBDD), medicinal chemistry applications. We discuss how these technologies are facilitating remaining challenges identify binding sites ligands.

Language: Английский

---

Global atlas of predicted functional domains inLegionella pneumophilaDot/Icm translocated effectors

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

Published: May 9, 2024

Abstract Legionella pneumophila utilizes the Dot/Icm type IVB secretion system to deliver hundreds of effector proteins inside eukaryotic cells ensure intracellular replication. Our understanding molecular functions this largest pathogenic arsenal known bacterial world remains incomplete. By leveraging advancements in 3D protein structure prediction, we provide a comprehensive structural analysis 368 L. effectors, representing global atlas predicted functional domains summarized database ( https://pathogens3d.org/legionella-pneumophila ). identified 157 types diverse 287 including 159 effectors with no prior annotations. Furthermore, 35 unique 30 models that have similarity experimentally structurally characterized proteins, thus, hinting at novel functionalities. Using analysis, demonstrate activity thirteen domains, three folds, cause growth defects Saccharomyces cerevisiae model system. This illustrates an emerging strategy exploring synergies between predictions and targeted experimental approaches elucidating activities involved infection.

Language: Английский

---

Integration of co-culture and transport engineering for enhanced metabolite production

Plant Biotechnology, Journal Year: 2024, Volume and Issue: 41(3), P. 195 - 202

Published: Aug. 6, 2024

Microbial production of valuable plant metabolites is feasible. However, constructing all pathways in a single cell formidable challenge, and the extended biosynthetic within cells often result reduced productivity. To address these challenges, co-culture system that divides into several host co-cultures has been developed. Various combinations cells, along with optimal conditions for each co-culture, have documented, leading to successful metabolites. In addition, efficient biosynthesis frequently involves metabolite movement, encompassing substrate uptake, intracellular intermediate transport, end-product efflux. Recent advances transporters specialized enhanced productivity by harnessing transporters. This review summarizes latest findings on systems transport engineering provides insights future through integration engineering.

Language: Английский

---

FIBOS: R and Python packages for analyzing protein packing and structure

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

Published: Nov. 3, 2024

Abstract Summary Advances in prediction of the 3D structures most known proteins through machine learning methods have achieved unprecedented accuracies. However, although these computed structure models (CSMs) are overall remarkably good, they still challenge accuracy at atomic level, especially refinement interatomic packing. The Occluded Surface (OS) algorithm is widely used for packing analysis. But it lacks implementations high-level languages like R and Python, limiting its accessibility. We introduce FIBOS, an Python package incorporating OS methodology with enhancements. It embeds efficient Fortran code from original OS. also improves by using Fibonacci spirals to distribute surface dots, reducing anisotropy ensuring a more even point distribution. As case study we compared densities between experimental protein AlphaFold predictions. While average similar, exhibit comparatively greater variability less homogeneity, potentially reflecting local inaccuracies. FIBOS provides accessible tool density analysis facilitating structural assessment era extensive computational models. Availability implementation packages available https://github.com/insilico-unifei/fibos-R https://github.com/insilico-unifei/fibos-py .

Language: Английский

---