The trihelix transcription factor family, also known as GT factors, plays a key role in regulating plant growth, development, and responses to various abiotic stresses. These plant‐specific transcription factors are characterized by a conserved trihelix DNA‐binding domain. This study presents the first comprehensive genome‐wide identification and bioinformatic analysis of Trihelix genes in common bean. A total of 38 Trihelix genes ( PvTH ) were identified and classified into five subgroups: GT‐1, GT‐2, GT γ , SH4, and SIP1. Analyses included phylogenetic relationships, chromosomal distribution, tandem and segmental duplication events, gene structures, promoter regions, exon–intron organization, miRNA interactions, conserved motif identification, 3D protein structure predictions, and synteny relationships with Arabidopsis and soybean. Additionally, RNA‐seq data were used to evaluate the expression profiles of PvTH genes under salt and H 2 O 2 stress, and qPCR analyses confirmed the altered mRNA expression levels after stress treatments. Several PvTH genes showed statistically significant responses under these stress conditions, suggesting their involvement in abiotic stress responses in beans. The findings of this study are novel and contribute to the understanding of the Trihelix gene family in common bean, providing a valuable resource for further functional and stress‐related research in legume crops.

From sequence to structure: A comprehensive review of deep learning models for RNA structure prediction.

Intrinsically disordered proteins are ubiquitous in biological systems and play essential roles in a wide range of biological processes and diseases. Despite recent advances in high-resolution structural biology techniques and breakthroughs in deep learning-based protein structure prediction, accurately determining structural ensembles of IDPs at atomic resolution remains a major challenge. Here, we introduce bAIes, a Bayesian framework that integrates AlphaFold2 predictions with physico-chemical molecular mechanics force fields to generate accurate atomic-resolution ensembles of IDPs. We show that bAIes produces structural ensembles that match a wide range of high- and low-resolution experimental data across diverse systems, achieving accuracy comparable to atomistic molecular dynamics simulations but at a fraction of their computational cost. Furthermore, bAIes outperforms state-of-the-art IDP models based on coarse-grained potentials as well as deep-learning approaches. Our findings pave the way for integrating structural information from modern deep-learning approaches with molecular simulations, advancing ensemble-based understanding of disordered proteins.

Hypokalemic periodic paralysis (HypoPP) is a muscle disease caused by abnormal ion channels and is characterized by recurrent skeletal muscle relaxation paralysis and hypokalemia. There are obvious triggers before disease onset, such as cold, excessive exercise, excessive consumption of sugary and high-energy foods, and overeating. The aim of this study was to elucidate the pathogenic mechanism of novel mutations in the voltage-dependent L-type calcium channel subunit alpha-1S (CACNA1S) gene associated with HypoPP. Method: Whole-exome sequencing and American College of Medical Genetics and Genomics (ACMG) compliance analysis were performed, supplemented by serum potassium and blood biochemistry tests for bioinformatics analysis. We report a 13-year-old adolescent male patient with hypokalemic periodic paralysis, who complained of limb muscle weakness accompanied by pain for 10h. Whole-exome sequencing revealed a mutation in the CACNA1S gene (NM_000069.3: exon27: c.3491A>C [p. Glu1164Ala]), which was classified as an uncertain mutation. The clinical presentation and protein structure prediction of the gene mutation confirmed its pathogenic role and mechanism. The mutation caused a conformational change in the calcium ion channel. This study revealed a new mutation site in the HypoPP gene and proposed the possibility of a new pathogenesis. Moreover, obesity and low magnesium are two factors that induce HypoPP, which may increase the risk of disease.

Proteolysis Targeting Chimeras (PROTACs) represent a transformative approach to drug development by leveraging the intracellular ubiquitin-proteasome system (UPS) for the selective degradation of target proteins. A PROTAC molecule comprises three essential components: a ligand that binds to the E3 ubiquitin ligase, a ligand that targets the protein of interest, and a flexible linker that connects the two. This distinctive structure enables the PROTAC to simultaneously engage with both the target protein and the E3 ligase, facilitating their interaction. Such proximity initiates the ubiquitination of the target protein, marking it for recognition and subsequent degradation. In this study, we benchmark ternary complexes based on PROTACs using four recently employed predictive tools: Chai-1, AlphaFold2, AlphaFold3, and Protenix. Comparative analysis indicated that the ternary complexes predicted by the four prediction tools demonstrated satisfactory accuracy (Cα-RMSD < 10 Å). Among the evaluated tools, three-Chai-1, AlphaFold3, and Protenix-demonstrated superior performance in over half of the tests, while AlphaFold2 exhibited comparatively lower performance. However, significant challenges remained in accurately predicting the orientation of POI and the E3 ligase (Cα-RMSD < 10 Å when POI or E3 ligase were used as reference), as well as the position of the small molecule PROTAC (RMSD < 5 Å). By benchmarking these tools, we underscore recent advancements in protein structure prediction, enhance our understanding of the mechanisms underpinning PROTAC complexes, and provide a valuable reference for evaluating the binding conformations of other ternary complexes, as well as for the development of future predictive tools.

Protein structure bridges the sequence-function relationship, enabling deep exploration of biological processes across diverse organisms. Insects, the most diverse animal lineage, accounting for over 50% of all described animal species, provide an exceptional system for exploring sequence-structure-function relationships. Here, we reconstructed a comprehensive and well-resolved phylogeny of 4854 insects, spanning all orders. Leveraging this framework, we created an atlas of 13.29 million predicted protein structures from 824 representative species, including 11.63 million newly predicted structures. Structural clustering revealed that proteins with divergent sequences but similar structures could be effectively grouped together. Structural similarity searches against proteins with well-characterized functions yielded annotations for 7.61 million insect proteins, including up to 14% of previously unannotated proteins. We further identified 750 million remote homologs between insect proteins, many of which trace back to ancient branches of the insect phylogeny. Remarkably, despite extensive sequence divergence, cGAS-like receptors (cGLRs) were structurally conserved across all 824 insects. Experimental assays demonstrated that these structurally identified cGLRs play a crucial role in antiviral defense in the yellow fever mosquito. Our findings highlight the significance of structural genomics for understanding protein function and evolution across the tree of life.

Helianthus annuus L., as one of the important oil crops, has strong salt and drought tolerance. The RNA-dependent RNA polymerase (RDR) plays an irreplaceable role in plant growth and development and in the formation of small-molecule RNA. To clarify the functions and regulatory mechanisms of this gene family, in this study, bioinformatics analysis was performed to identify the members of the RDR gene family in H. annuus, and the physicochemical properties, chromosome location, phylogenetic relationship, and subcellular localization of the family members were analyzed in depth. Meanwhile, RT-qPCR was conducted to explore the expression patterns of the family members under salt, alkali, and drought stresses. The results showed that a total of 10 members of the RDR gene family were identified in H. annuus, and they were distributed on chromosomes 1, 3, 6, 8, and 16. The phylogenetic analysis indicated that the RDRs of H. annuus and Arabidopsis were clustered into 4 groups. The results of multiple amino acid sequence alignment showed that all the members of this family contained the conserved domain of RdRP. Protein structure prediction revealed that the secondary structure of the protein family members was mainly composed of α-helixes and random coils. STRING interaction network analysis revealed that HaRDR interacted with Argonaute (HaAGO) and Dicer-like (HaDCL) proteins. The results of RT-qPCR revealed that the expression levels of the HaRDR gene family members were significantly upregulated in stems under salt stress and in leaves under alkali stress. The subcellular localization results indicated that HaRDR3c was located in the cytoplasm. The results of this study suggest that the HaRDR family members are highly conserved during evolution while exhibiting functional diversity. The RT-qPCR results indicate that the HaRDR family members can respond to abiotic stresses. The findings not only provide a basis for exploring the role of HaRDR in regulating stress resistance but also lays a foundation for revealing the molecular mechanism of HaRDR in plant stress responses.

Artificial intelligence (AI) models have advanced rapidly, driving breakthroughs in protein structure prediction, functional annotation, and conformational exploration. Among these, molecular dynamics (MD)-inspired generative models such as AlphaFlow and BioEmu show strong potential for capturing conformational ensembles. In this study, we benchmark these models alongside physics-based MD simulations to evaluate their ability to detect cryptic pockets in proteins. Identifying such transient pockets remains a vital goal in drug discovery, as they can offer new avenues for targeting proteins traditionally challenging to modulate. We also assess two specialized residue-level predictors, PocketMiner and CryptoBank. Using the interferon inhibitory domain of Zaire Ebola VP35 (VP35), TEM-1 β-lactamse with the M182T substitution (TEM β-lactamase), and their mutants, we test whether each method can detect pockets and capture the effects of point mutations known to enhance or suppress pocket formation. All methods successfully identify pockets in VP35 and distinguish between opening and closing mutants. However, in TEM, where pocket opening is subtle, the methods perform inconsistently. These results highlight the promise of AI-based and simulation-based strategies in cryptic pocket discovery while pointing to the need for further improvements to achieve robust, system-independent predictions.

Soybean is a globally important economic and food crop, whose production is often constrained by drought stress, posing a serious threat to yield and quality. Genomic selection (GS) has become a core technology in modern breeding, effectively enhancing breeding efficiency. However, conventional prediction models mainly rely on additive genetic effects and fail to adequately incorporate non-additive factors such as epistasis, limiting further improvements in prediction accuracy. In this study, a genome-wide epistatic analysis of soybean drought tolerance identified 3594 protective interaction pairs. Incorporating significant epistatic SNP pairs into six genomic prediction models resulted in comparable and substantial improvements in prediction accuracy across all models (by 24%). Furthermore, integration of AlphaFold2-based protein structure prediction and transcriptional regulatory analyses validated the biological reliability of protective epistatic pairs, effectively reducing the risk of false positives. Network construction and functional enrichment analyses further revealed that these epistatic pairs participate in coordinated protein structural interactions and are enriched in key biological pathways. Haplotype analysis confirmed the critical regulatory role of non-additive effects in soybean drought tolerance. Collectively, this study establishes a comprehensive evidence chain from molecular mechanisms to breeding applications, demonstrating that integrating epistasis into GS can effectively enhance prediction performance for drought tolerance in soybean. These findings provide novel research strategies for the genetic analysis of complex traits and efficient breeding.

The rational design of chemical compounds that bind to a desired protein target molecule is a major goal of drug discovery. Most current molecular docking but also fragment-based build-up or machine-learning based generative drug design approaches employ a rigid protein target structure. Based on recent progress in predicting protein structures and complexes with chemical compounds we have designed an approach, AI-MCLig, to optimize a chemical compound bound to a fully flexible and conformationally adaptable protein binding region. During a Monte-Carlo (MC) type simulation to randomly change a chemical compound the target protein-compound complex is completely rebuilt at every MC step using the Chai-1 protein structure prediction program. Besides compound flexibility it allows the protein to adapt to the chemically changing compound. MC-protocols based on atom/bond type changes or based on combining larger chemical fragments have been tested. Simulations on three test targets resulted in potential ligands that show very good binding scores comparable to experimentally known binders using several different scoring schemes. The MC-based compound design approach is complementary to existing approaches and could help for the rapid design of putative binders including induced fit of the protein target. Datasets, examples and source code are available on our public GitHub repository https:/github.com/JakobAgamia/AI-MCLig and on Zenodo at https://doi.org/10.5281/zenodo.17800140.

Deep learning-based methods, such as AlphaFold2, have revolutionized the prediction of static protein structures. However, modeling alternative conformations and dynamic structures remains an unsolved problem. Here, we present trRosettaX2-Dynamics (trX2-D), an innovative solution building on our CASP15 and CASP16 winning method, trRosettaX2. trX2-D tackles this challenge by employing physics-based iterative sampling of trRosettaX2's predicted inter-residue geometric distributions. The model underwent pre-training on high-resolution X-ray structures, followed by fine-tuning on approximately 7000 dynamic NMR structures. This dual training regime significantly bolsters its capacity to predict alternative conformations and dynamic structures. At its core, trX2-D employs a Transformer-based neural network to initially predict a set of inter-residue geometric constraints. These constraints are then iteratively sampled to generate dynamic structures, entirely circumventing the need for prior knowledge of native structural states. Extensive benchmarking across three distinct datasets-two focused on alternative conformations and one on dynamic structures-demonstrates trX2-D's promising ability to predict alternative conformations and accurately capture structural dynamics. This work highlights the potential of integrating deep learning predictions with physics-based sampling to advance the field of protein dynamic structure prediction.

Folding stability is crucial for the vast majority of proteins. Computational methods suggested to date for the absolute folding stability (ΔG) prediction, including those driven from protein structure prediction AIs, show clear limitations in reproducing quantitative experimental values. Here we present IFUM, a deep neural network that jointly estimates ΔG and the equilibrium ensemble of folded and unfolded states represented by residue-pair distance probability distributions. This joint learning considerably enhances prediction accuracy compared to learning ΔG alone. Trained on a dataset including Mega-scale small proteins, disordered proteins, and wild-type natural proteins, IFUM is robust to various protein types and can accurately predict complex mutational effects like insertions or deletions. Here, we show that IFUM effectively guides real-world design challenges, exhibiting strong correlation with experimental melting temperatures in protein engineering and outperforming AlphaFold-based metrics in de novo design selection.

Atkinsonella hypoxylon virus (AhV) is a fungi-infecting betapartitivirus and the typical member of the Partitiviridae, a family of persistent viruses that infect a broad range of organisms. Partitiviruses have been largely overlooked following their designation as cryptic viruses. However, evidence is accumulating that they play an important role in the ecology of their hosts. Since the capsid proteins of partitiviruses have been implicated in virus–host interactions, exploring their structural biology may give clues into the evolution, horizontal transmission and host adaptation of partitiviruses. The capsid of AhV shares the same organization of other partitiviruses with 60 dimeric capsid protein protomers arranged with T=1 icosahedral symmetry. The structure, determined by cryo-electron microscopy to 2.4 Å, shows that AhV has a unique iteration on the protrusion domain with an extensive network of hydrophobic interactions among equivalent interdigitating loops at the dimerization interface. AhV also shares a conserved helical core in the shell domain, which we extend to all genera of the recognized partitiviruses using protein structure prediction. The helical core appears to be a conserved element of the picobirnavirus lineage of capsid protein folds and provides a template onto which various elaborations of the protrusion domain have evolved. The involvement of the protrusion in virus–host interactions has previously been proposed, and our findings provide evidence of a structural device enabling capsid protein diversification during the evolution of the Partitiviridae.

This systematic review outlines pivotal advancements in deep learning-driven protein structure prediction and design, focusing on four core models—AlphaFold, RoseTTAFold, RFDiffusion, and ProteinMPNN—developed by 2024 Nobel Laureates in Chemistry: David Baker, Demis Hassabis, and John Jumper. We analyze their technological iterations and collaborative design paradigms, emphasizing breakthroughs in atomic-level structural accuracy, functional protein engineering, and modeling multi-component biomolecular interactions. Key innovations include AlphaFold3's diffusion-based framework for unified biomolecular prediction, RoseTTAFold's three-track architecture integrating sequence and spatial constraints, RFDiffusion's denoising diffusion for de novo protein generation, and ProteinMPNN's inverse folding for sequence–structure co-optimization. Despite major progress in applications such as binder design, nanomaterials, and enzyme engineering, challenges persist in dynamic conformational sampling, multimodal data integration, and generalization to non-canonical targets. We propose future directions, including hybrid physics-AI frameworks and multimodal learning, to bridge gaps between computational design and functional validation in cellular environments.

Extracellular vesicles (EVs) are emerging as promising noninvasive biomarkers, yet their clinical translation faces substantial hurdles, primarily due to the challenge of identifying assay-compatible markers. Here, in this Review, we outline sophisticated computational frameworks, particularly leveraging artificial intelligence, to bridge this gap. We detail the integration of diverse data resources, including disease-specific omics, EV, protein localization, tissue-specific, drug, model system and immune databases. This Review comprehensively describes computational selection strategies, from rule-based sequential filtering to advanced machine learning for data fusion and deep learning for multi-omics integration. Crucially, it discusses the refinement of biomarker candidates using artificial-intelligence-driven predictions of protein structure and physicochemical properties, ensuring compatibility with existing assay systems. By systematically evaluating biomarkers for predictive performance, biological plausibility and clinical utility, this framework aims to accelerate the transition of EV research from discovery to clinical application, thereby enhancing precision medicine.

Understanding how protein structures dictate their diverse biological functions remains one of the central and enduring challenges in structural biology. The development of AlphaFold and ESMAtlas marks a significant advance in protein science, enabling the reliable prediction of protein structure directly from amino acid sequence. This advance in structure prediction underscores the need for complementary methods that can explore conformational space and enable efficient sampling of dynamic trajectories. Here, we present TSS-Pro, a conditional generative diffusion framework that enables efficient sampling of protein conformational trajectory space. TSS-Pro takes the initial frame as conditional input and generates protein conformational trajectories. It supports two sampling strategies: (1) consecutive sampling, where each trajectory segment is generated step by step by conditioning on the final frame of the previously generated segment, enabling temporally coherent propagation of structural transitions; (2) parallel sampling, where multiple trajectory branches are independently generated from initial conditions to enhance conformational diversity. We validate TSS-Pro on three representative systems of increasing complexity: alanine dipeptide, ubiquitin, and Drosophila cryptochrome (dCRY). TSS-Pro reproduces the free energy landscape of alanine dipeptide. In the case of ubiquitin, consecutive sampling with TSS-Pro overcomes local minima and uncovers distinct conformational states of the C-terminal region. For the large protein dCRY, TSS-Pro achieves high efficiency through parallel trajectory sampling, enabling conformational and dynamic exploration typically accessible only through extensive simulations. TSS-Pro paves the way for high-throughput exploration of protein trajectories and conformational landscapes for large and complex systems.

The diverse regulatory functions, protein production capacity, and stability of natural and synthetic RNAs are closely tied to their ability to fold into intricate structures. Determining RNA structure is thus fundamental to RNA biology and bioengineering. Among existing approaches to structure determination, computational secondary structure prediction offers a rapid and low-cost strategy and is thus widely used, especially when seeking to identify functional RNA elements in large transcriptomes or screen massive libraries of novel designs. While traditional approaches rely on detailed measurements of folding energetics and/or probabilistic modeling of structural data, recent years have witnessed a surge in deep learning methods, inspired by their tremendous success in protein structure prediction. However, the limited diversity and volume of known RNA structures can impede their ability to accurately predict structures markedly different from the ones they have seen. This is known as the generalization gap and currently poses a major barrier to progress in the field. In this Perspective article, we gauge method generalizability using a new benchmark dataset of structured RNAs we curated from the Protein Data Bank. We also discuss the emergence of deep learning methods for predicting structure probing data and use a new dataset to underscore generalization challenges unique to this domain along with directions for future improvement. Expanding beyond improving predictive accuracy, we review how advances in deep learning have recently enabled scalable and accessible optimization of traditional structure prediction methods and their seamless integration with modern neural networks.

Recent breakthroughs in protein structure prediction have opened new avenues for genome-wide drug discovery, yet existing virtual screening methods remain computationally prohibitive. We present DrugCLIP, a contrastive learning framework that achieves ultrafast and accurate virtual screening, up to 10 million times faster than docking, while consistently outperforming various baselines on in silico benchmarks. In wet-lab validations, DrugCLIP achieved a 15% hit rate for norepinephrine transporter, and structures of two identified inhibitors were determined in complex with the target protein. For thyroid hormone receptor interactor 12, a target that lacks holo structures and small-molecule binders, DrugCLIP achieved a 17.5% hit rate using only AlphaFold2-predicted structures. Finally, we released GenomeScreenDB, an open-access database providing precomputed results for ~10,000 human proteins screened against 500 million compounds, pioneering a drug discovery paradigm in the post-AlphaFold era.

A protein's function depends critically on its conformational ensemble, a collection of energy weighted structures whose balance depends on temperature and environment. Though recent deep learning (DL) methods have substantially advanced predictions of single protein structures, computationally modeling conformational ensembles remains a challenge. Here, we focus on modeling fold-switching proteins, which remodel their secondary and/or tertiary structures and change their functions in response to cellular stimuli. These underrepresented members of the protein universe serve as test cases for a method's generalizability. They reveal that DL models often predict conformational ensembles by association with training-set structures, limiting generalizability. These observations suggest use cases for when DL methods will likely succeed or fail. Developing computational methods that successfully identify new fold-switching proteins from large pools of candidates may advance modeling conformational ensembles more broadly.

Soluble sugars are the key photo-assimilates in higher plants, playing critical roles in growth, development, and stress regulation. The transport of sugars in plants involves the coordinated action between several sugar transporter families, including the SUT, STP, pGlcT, VGT, TMT, INT, PLT, SFP, and SWEET families. Over recent decades, numerous studies have elucidated the molecular functions of major sugar transporters. Phylogenetic and evolutionary analyses support the conservation of substrate specificity and transport direction, at least to some extent. Structural analyses have provided key insights into the structural-function relationships of important transporters (e.g., OsSWEET2b and AtSTP10), which can be effectively leveraged for artificial intelligence (AI)-enabled protein structure prediction and rational design. Advances in omics technologies now enable low-cost, routine transcriptome profiling and cutting-edge techniques (e.g., single-cell multi-omics and spatiotemporal RNA-seq), providing unprecedented ways to understand how sugar transporters function coordinately at multiple levels. Here, we describe the classification of major sugar transporters in plants and summarize established functional knowledge. We emphasize that recent groundbreaking advances in AI-enabled protein analyses and multi-omics will revolutionize molecular physiology in crops. Specifically, the integration of functional knowledge, AI-based protein analyses, and multi-omics will help unravel the orchestration of different sugar transporters, thereby enhancing our understanding of how sugar transportation and source-sink interactions contribute to crop development, yield formation, and beyond, ultimately boosting carbohydrate transport- related crop improvement.