Biology subjects

Zheng, L.

Publications and source records attributed to Zheng, L..

At least 109 records · Page 6Linked to original sources

bioRxiv2024

A multiplexed and quantitative assay for small-molecule detection via photo-crosslinking of structure switching aptamers

There is an unmet need for molecular detection assays that enable the multiplexed, quantification of smallmolecule analytes. We present xPlex, an assay that combines aptamer switches with ultraviolet-crosslinkable complementary strands to record target-binding events. When the aptamers small-molecule target is present, the crosslinkable strand is displaced, enabling PCR amplification and detection of the relevant aptamer. In the absence of that target, the aptamer is readily cross-linked to the strand, preventing amplification from happening. The resulting aptamer-specific amplicons can be detected and quantified in a multiplexed fashion using high-throughput sequencing. We demonstrate quantitative performance for a pair of small-molecule analytes, dopamine and glucose, and show that this assay retains good specificity with mixtures of the two molecules at various concentrations. We further show that xPlex can effectively evaluate the specificity of cross-reactive aptamers to a range of different small-molecule analytes. We believe the xPlex assay format could offer a useful strategy for achieving multiplexed analysis of small-molecule targets in a variety of scenarios.

bioengineering↗

bioRxiv2024

Exploring Gene-Mediated Mechanisms Behind Shared Phenotypes Across Diverse Diseases Using the clGENE Tool

The observation of similar clinical characteristics across a broad spectrum of diseases suggests the existence of underlying shared molecular mechanisms. Identifying these mechanisms is critical for uncovering the molecular roots of diseases and advancing the development of innovative therapeutic strategies. However, researching the common genes that mediate similar phenotypes among different diseases often requires the integration of various sequencing datasets and clinical data. The batch effects among these datasets and the complexity of clinical data present significant challenges to the research. This study developed a framework named "clGENE", aimed at uncovering the molecular mechanisms behind similar phenotypes across different diseases. By integrating data normalization, cosine similarity analysis, and principal component analysis (PCA) algorithms, this framework is capable of effectively identifying shared molecular mechanisms associated with specific phenotypes and further selecting key shared genes. Through the analysis of a pan-cancer dataset, we have verified the efficacy and reliability of the "clGENE" framework. Furthermore, this study also established a dataset on immune cell infiltration and successfully identified key patterns of immune cell infiltration in different cancer lymph node metastasis stages using the clGENE framework, further confirming its potential application in biomedical research.

bioinformatics↗

bioRxiv2024

The subcommissural organ regulates brain development via secreted peptides

The subcommissural organ (SCO) is a gland located at the entrance of the aqueduct of Sylvius in the brain. It exists in species as distantly related as amphioxus and humans, but its function is largely unknown. To explore its function, we compared transcriptomes of SCO and non-SCO brain regions and found three genes, Sspo, Car3, and Spdef, that are highly expressed in the SCO. Mouse strains expressing Cre recombinase from endogenous promoter/enhancer elements of these genes were used to genetically ablate SCO cells during embryonic development, resulting in severe hydrocephalus and defects in neuronal migration and development of neuronal axons and dendrites. Unbiased peptidomic analysis revealed enrichment of three SCO-derived peptides, namely thymosin beta 4, thymosin beta 10, and NP24, and their reintroduction into SCO-ablated brain ventricles substantially rescued developmental defects. Together, these data identify a critical role for the SCO in brain development.

neuroscience↗

bioRxiv2024

Map based cloning of CT2 and the pilot functional exploration in abiotic stress

Heterotrimeric G-proteins are multifunctional modulators that participate in a wide range of growth and developmental processes in eukaryotic species, from yeast to plants and animals. Component detection and the study of G protein signaling in most plants, including maize, are in the initial stages. In this study, we characterized a maize mutant, ct2, that showed a compact architecture and reproductive organ-related phenotypic variation. The target gene CT2 was cloned using bulked segregant analysis and map-based cloning. Gene structure prediction and phylogenetic analysis indicated that CT2 is a canonical G protein belonging to the monocotyledonous group. Promoter analysis of CT2 and RNA sequencing revealed cis-acting regulatory elements and differentially expressed genes involved in JA signaling and stress tolerance. The transcription of CT2 was repressed by NaCl and PEG treatments, and ct2 mutation in the ct2 line compromised stress tolerance in maize. On the basis of our results, we proposed a schema diagram of CT2-regulated biological process and their feedback on CT2 transcription. This research provides clues for further studies of CT2 function in hormone signaling and stress tolerance, which is beneficial for maize breeding through the screening and application of beneficial alleles.

molecular biology↗

bioRxiv2024

Why the adventitious roots of poplar are so colorful: RNAseq and metabolomic analysis reveal flavonols, flavones, and anthocyanins accumulation in canker pathogens-induced adventitious roots in poplar

Recently, we observed a novel allometry on poplar stems, with copious colorful adventitious roots (ARs) induced by fungal canker pathogens. Here, we reveal chemical, physiological, and molecular mechanisms of AR coloration in a poplar-pathogen (Valsa sordida/Botrosphaeria dothidea) interaction system using our phloem girdling-inoculation system. Light-induced coloration in ARs: red/rosy under sunlight and milky white under shading. Chemical and metabolomic analyses indicated that numerous (93 in all 110) and high relative intensities/contents of flavonoids metabolites (mainly including flavonols, flavones, and anthocyanins class) accumulate in red ARs, some flavones and anthocyanins metabolites all contribute to the color of poplar ARs, and cyanidin-3-O-glucoside is the most abundant colorant. Integrated analysis of metabolomic and transcriptomic analysis suggested that sunlight exposure redirected metabolomic flux from the flavonoid biosynthesis pathway to the flavonols and flavones branch pathways, induced by the upregulation of FLS (flavonol synthase/flavanone 3-hydroxylase) and other structural genes. The anthocyanins metabolomic analysis and the downregulation of the ANS (anthocyanin synthase) gene illustrated a retard of metabolomic flux from leucoanthocyanidins to anthocyanidins; meanwhile, metabolomic results and the upregulation of gene BZ1 (Bronze 1, anthocyanin 3-O-glucosyltransferase) illustrated that sunlight triggered a rapid biosynthesis of anthocyanin metabolites in poplar ARs, which based on the substrate level of anthocyanidins. Transcriptomic and RT-qPCR analyses showed that transcriptional factor MYB113, HY5 (ELONGATED HYPOCOTYL5), and COP1 (Ring-finger protein CONSTITUTIVE PHOTOMORPHOGENIC1) genes positively regulate the expression of the flavonoid/anthocyanin biosynthesis structural genes (such as genes encoding BZ1, FLS, LAR, etc.) in both sunlight-exposed red ARs and white ARs after light exposure, suggesting sunlight induces anthocyanins biosynthesis through the interaction between "MBW" complex and COP1-HY5 module. Moreover, results also showed that 1 SPL gene (squamosa promoter-binding-like protein gene, target of miR156), one component of miR156-SPL module, downregulated in sunlight-exposed poplar ARs, implying the biosynthesis flavonoid/anthocyanin be regulated at the posttranscriptional level. Additionally, this study provides a potential AR experimental system for research on flavonoid/anthocyanin biosynthesis in tree species.

plant biology↗

bioRxiv2024

A computational and multi-brain signature for aberrant social coordination in schizophrenia

Social functioning impairment is a core symptom of schizophrenia (SCZ). Yet, the computational and neural mechanisms of social coordination in SCZ under real-time and naturalistic settings are poorly understood. Here, we instructed patients with SCZ to coordinate with a healthy control (HC) in a joint finger-tapping task, during which their brain activity was measured by functional near-infrared spectroscopy simultaneously. The results showed that patients with SCZ exhibited poor rhythm control ability and unstable tapping behaviour, which weakened their interpersonal synchronization when coordinating with HCs. Moreover, the dynamical systems modelling revealed disrupted between-participant coupling when SCZ patients coordinated with HCs. Importantly, increased inter-brain synchronization was identified within SCZ-HC dyads, which positively correlated with behavioural synchronization and successfully predicted dimensions of psychopathology. Our study suggests that SCZ individuals may require stronger neural alignment to compensate for deficiency in their coordination ability. This hyperalignment may be relevant for developing inter-personalized treatment strategies.

neuroscience↗

bioRxiv2024

Thymidine Phosphorylase Mediates SARS-CoV-2 Spike Protein Enhanced Thrombosis in K18-hACE2TG Mice

COVID-19, caused by SARS-CoV-2, is associated with arterial and venous thrombosis, thereby increasing mortality. SARS-CoV-2 spike protein (SP), a viral envelope structural protein, is implicated in COVID-19-associated thrombosis. However, the underlying mechanisms remain unknown. Thymidine phosphorylase (TYMP), a newly identified prothrombotic protein, is upregulated in the plasma, platelets, and lungs of patients with COVID-19 but its role in COVID-19-associated thrombosis is not defined. In this study, we found that wild-type SARS-CoV-2 SP significantly promoted arterial thrombosis in K18-hACE2TG mice. SP-accelerated thrombosis was attenuated by inhibition or genetic ablation of TYMP. SP increased the expression of TYMP, resulting in the activation of signal transducer and activator of transcription 3 (STAT3) in BEAS-2B cells, a human bronchial epithelial cell line. A siRNA-mediated knockdown of TYMP inhibited SP-enhanced activation of STAT3. Platelets derived from SP-treated K18-hACE2TG mice also showed increased STAT3 activation, which was reduced by TYMP deficiency. Activated STAT3 is known to potentiate glycoprotein VI signaling in platelets. While SP did not influence ADP- or collagen-induced platelet aggregation, it significantly shortened activated partial thromboplastin time and this change was reversed by TYMP knockout. Additionally, platelet factor 4 (PF4) interacts with SP, which also complexes with TYMP. TYMP enhanced the formation of the SP/PF4 complex, which may potentially augment the prothrombotic and procoagulant effects of PF4. We conclude that SP upregulates TYMP expression, and TYMP inhibition or knockout mitigates SP-enhanced thrombosis. These findings indicate that inhibition of TYMP may be a novel therapeutic strategy for COVID-19-associated thrombosis. Key PointsO_LISARS-CoV-2 spike protein, thymidine phosphorylase, and platelet factor 4 form a complex that may promote clot formation. C_LIO_LIInhibiting thymidine phosphorylase attenuates SARS-CoV-2 spike protein-enhanced thrombosis, platelet activation, and coagulation. C_LI

cell biology↗

bioRxiv2024

The MKK3 MAPK cascade integrates temperature and after-ripening signals to modulate seed germination

Temperature is a major environmental cue for seed germination. The permissive temperature range for germination is narrow in dormant seeds and expands during after-ripening. Quantitative trait loci analyses of pre-harvest sprouting in cereals have revealed that MKK3, a mitogen-activated protein kinase (MAPK) cascade protein, is a negative regulator of grain dormancy. Here we show that the MAPKKK19/20-MKK3-MPK1/2/7/14 cascade modulates germination temperature range in Arabidopsis seeds by elevating germinability of the seeds at sub- and supra-optimal temperatures. The expression of MAPKKK19 and MAPKKK20 is regulated by an unidentified temperature sensing and signaling mechanism the sensitivity of which is modulated during after-ripening of the seeds, and MPK7 is activated at the permissive temperature for germination regulated by expression levels of MAPKKK19/20. Activation of the MKK3 cascade represses abscisic acid (ABA) biosynthesis enzyme gene expression, and induces expression of ABA catabolic enzyme and gibberellic acid biosynthesis enzyme genes, resulting in expansion of the germinable temperature range. Our data demonstrate that the MKK3 cascade integrates temperature and after-ripening signals to germination processes including phytohormone metabolism.

plant biology↗

bioRxiv2024

Turnover of RNA-binding Proteins and MicroRNAs by intrinsically disordered region-directed ZSWIM8 ubiquitin ligase during brain development

Widely present in mammalian proteomes, intrinsically disordered regions (IDRs) in proteins play important biological functions by conferring structural flexibility and mediating biomolecular interactions. IDR-containing proteins, including many RNA-binding proteins (RBPs), are prone to misfolding and aggregation and must be constantly monitored. Here we show that the conserved ZSWIM8-type Cullin-RING ubiquitin ligase (CRLZSWIM8) is a master regulator of such proteins during brain development. ZSWIM8 selects its substrates via an IDR-dependent mechanism, and deletion of ZSWIM8 causes aberrant accumulation of numerous RBPs including AGO2 and ELAV1 in neonatal brains. Furthermore, AGO2 ubiquitination by ZSWIM8 is triggered by microRNA binding, leading to target-directed microRNA degradation (TDMD) of MiR7. Dysregulation of MiR7 in the absence of ZSWIM8 results in defects in oligodendrocyte maturation and functions. Together, our findings have demonstrated that, by utilizing variable target-recognition strategies, ZSWIM8 controls the abundance of conformationally flexible RBPs and miRNA metabolism that are essential for brain development. TeaserA conserved ubiquitin ligase controls the quality of disordered proteins to ensure brain development.

developmental biology↗

bioRxiv2024

An Evaluation of the Tumor Microenvironment through CALR, IL1R1, IFNB1, and IFNG to Assess Prognosis and Immunotherapy Response in Bladder Cancer Patients

Immunogenic cell death (ICD) is a type of cell death sparking adaptive immune responses, can reshape the tumor microenvironment (TME). Exploring key ICD-related genes in bladder cancer (BLCA) could enhance personalized treatment. TCGA BLCA patients were divided into two ICD subtypes: ICD-high and ICD-low. High ICD expression linked to increased immune cell infiltration and longer survival, but with potentially suppressed immune function. The high ICD group responded better to PD1-targeted therapy. A risk-scoring model with four ICD-related genes (CALR, IL1R1, IFNB1, IFNG) was validated across TCGA, GEO datasets, and tissue samples, showing higher risk-score correlated with weaker anti-tumor immune function, more tumor-promoting elements, lower immunotherapy response rates, and shorter patient survival.This study connects ICD-related genes to BLCA prognosis and immune infiltration, offering a vital tool for personalized treatment guidance.

immunology↗

bioRxiv2024

The Mechanism Insight into Bacterial Degradation of Pentachlorobiphenyl

Bacterial degradation mechanism for high chlorinated pentachlorobiphenyl (PentaCB) with worse biodegradability has not been fully elucidated, which could limit the full remediation of PCBs-combined pollution. In this research, using enzymatic screening method, a new PentaCB-degrading bacterium M. paraoxydans that has not been reported was obtained. The characteristic of its intracellular enzymes, proteome and metabolome variation during PentaCB degradation were investigated systematically. The results showed that PentaCB (PCB101, 1 mg/L) degradation rate could arrive 23.9% within 4 h till complete degradation within 12 h. The intracellular enzyme compound was optimally active at pH 6.0. The 12 up-regulated characterized proteins involved ABC transporter substrate-binding protein, translocase protein TatA and signal peptidase I (SPase I) indicated that functional proteins for PentaCB degradation were present both in the cytoplasm and outer surface of cytoplasmic membrane. There were also 5 differential metabolites strongly associated with above proteins in which the up-regulated 1, 2, 4-benzenetriol was enriched into the degradation pathways of benzoate, chlorocyclohexane, chlorobenzene and aminobenzoate. Bacterial degradation of PentaCB necessitates transmembrane transport, energy consumption, protein export, biofilm formation and quorum sensing. These findings hold significant theory and application value for PCBs biodegradation.

microbiology↗

bioRxiv2024

Hydrophobic bulk conservation and genetic code determine the pattern of prokaryotic genome organization

Prokaryotic genomes in general exhibit marked organizational asymmetry, and differ substantially in GC-skew, AT-skew, and GS-bias (gene strand bias). Despite enigmatic origins, these organizational features and genomic GC-content have been described to be widely associated with each other recently, providing an opportunity to probe the evolutionary mechanisms. By analyzing sequence data of 4,012 leading strands from all available representative genomes, we found that the unusual nucleotide usage of coding genes under low GC-content contributed to a shift between two organizational patterns in the context of mutational biases. Analysis of artificially established neutral and natural models further suggests that in genes both GC-skew and AT-skew increase with decreasing GC-content, mostly contributed by synonymous substitutions because of genetic code, and by nonsynonymous substitutions because of hydrophobic bulk conservation in amino acid usage, respectively. This novel mechanistic framework in our study highlights the importance of evolutionary processes "operating at lower levels of organization" in the microbial world.

evolutionary biology↗

bioRxiv2023

Benchmarking Reverse Docking through AlphaFold2 Human Proteome

Predicting binding of a small molecule to the human proteome by reverse docking methods, we can predict the target interactions of drug compounds in the human body, as well as further evaluate their potential off-target effects or toxic side effects. In this study, we constructed 11 pipelines to evaluate and benchmark thoroughly the predictive capabilities of these reverse docking pipelines. The pipelines were built using site prediction tools (PointSite and SiteMap) based on the AF2 human proteome, docking programs (Glide and AutoDock Vina), and scoring functions (Glide, Autodock Vina, RTMScore, DeepRMSD, OnionNet-SFCT). The results show that pipeline glide_sfct (PS) exhibited the best target prediction ability and successfully predicted the similar proteins of native targets. This finding provides important clues for understanding the promiscuity between the drug ligand and the whole human proteome. In general, our study has the potential to increase the success rate and reduce the development timeline of drug discovery, thereby saving costs.

bioinformatics↗

bioRxiv2023

Fully flexible molecular alignment enables accurate ligand structure modelling

Accurate protein-ligand binding poses are the prerequisites of structure-based binding affinity prediction, and also provide the structural basis for in depth lead optimization in small molecule drug design. Ligand-based modeling approaches primarily extract valuable information from the structural features of small molecules to assess their potential as drug candidates against specific targets. However, it is challenging to provide reasonable predictions of binding poses for different molecules, due to the complexity and diversity of the chemical space of small molecules. Similarity-based molecular alignment techniques can effectively narrow the search range, as structurally similar molecules are likely to have similar binding modes, with higher similarity usually correlating to higher success rates. However, molecular similarity isnt consistently high because molecules often require changes to achieve specific purposes, leading to reduced alignment precision. To address this issue, we propose a new alignment method--Z-align. This method uses topological structural information as a criterion for evaluating similarity, reducing the reliance on molecular fingerprint similarity. Our method has achieved significantly higher success rates than other methods at moderate levels of similarity. Additionally, our approach can comprehensively and flexibly optimize bond lengths and angles of molecules, maintaining high accuracy even when dealing with larger molecules. Consequently, our proposed solution helps in achieving more accurate binding poses in protein-ligand docking problems, facilitating the development of small molecule drugs.

bioinformatics↗

bioRxiv2023

ThermoLink: Bridging Disulfide Bond and EnzymeThermostability through Database Construction andMachine Learning Prediction

Disulfide bonds (SS bonds), covalently formed by sulfur atoms in cysteine residues, play a crucial role in protein folding and structure stability. Due to their significance, artificial disulfide bonds are often introduced to enhance the thermostability of proteins. Although an increasing number of tools can assist with this task, significant amounts of time and resources were often wasted due to inadequate consideration. To enhance the accuracy and efficiency of designing disulfide bonds for protein thermostability improvement, we first collected disulfide-bond data with protein thermostability data from extensive literature sources. Then, various sequence- and structure-based features were extracted, and machine learning models were constructed to predict whether a disulfide bond could improve protein thermostability. Among all models, the neighborhood context model using the Adaboost-DT algorithm performs the best, and the AUC-ROC score and accuracy are 0.773 and 0.714, respectively. Alongside this, we also found that AlphaFold2 exhibits a high superiority in predicting disulfide bonds, and the coevolutionary relationship between residue pairs, to some extent, could also guide artificial disulfide-bond design. The SS-bond data has been integrated into an online server, named ThermoLink, available at guolab.mpu.edu.mo/thermoLink. O_TEXTBOXKey MessagesO_LIWe manually curated a database of disulfide bonds and their impacts on protein thermostability from the literature. C_LIO_LIAlphaFold2 exhibits a high superiority in predicting disulfide bonds, and to some extent, the conservation and co-evolutionary information of the residue pairs involved in disulfide bonds could also guide artificial disulfide-bond design. C_LIO_LIMachine learning models were developed to predict potential disulfide bonds with the aim of improving protein thermostability, offering an extended design strategy beginning with existing SS-bond prediction tools to optimize variant selection. C_LIO_LIThis work provides valuable data and theoretical support for a more precise design of disulfide bonds for improving protein thermostability. C_LI C_TEXTBOX

bioinformatics↗

bioRxiv2023

Semantical and Topological Protein Encoding Toward Enhanced Bioactivity and Thermostability

Protein engineering is a pivotal aspect of synthetic biology, involving the modification of amino acids within existing protein sequences to achieve novel or enhanced functionalities and physical properties. Accurate prediction of protein variant effects requires a thorough understanding of protein sequence, structure, and function. Deep learning methods have demonstrated remarkable performance in guiding protein modification for improved functionality. However, existing approaches predominantly rely on protein sequences, which face challenges in efficiently encoding the geometric aspects of amino acids local environment and often fall short in capturing crucial details related to protein folding stability, internal molecular interactions, and bio-functions. Furthermore, there lacks a fundamental evaluation for developed methods in predicting protein thermostability, although it is a key physical property that is frequently investigated in practice. To address these challenges, this paper introduces a novel pre-training framework that integrates sequential and geometric encoders for protein primary and tertiary structures. This framework guides mutation directions toward desired traits by simulating natural selection on wild-type proteins and evaluates variant effects based on their fitness to perform specific functions. We assess the proposed approach using three benchmarks comprising over 300 deep mutational scanning assays. The prediction results showcase exceptional performance across extensive experiments when compared to other zero-shot learning methods, all while maintaining a minimal cost in terms of trainable parameters. This study not only proposes an effective framework for more accurate and comprehensive predictions to facilitate efficient protein engineering, but also enhances the in silico assessment system for future deep learning models to better align with empirical requirements. The PyTorch implementation are available at https://github.com/tyang816/ProtSSN.

bioengineering↗

bioRxiv2023

Protein Engineering with Lightweight Graph Denoising Neural Networks

Protein engineering faces challenges in finding optimal mutants from the massive pool of candidate mutants. In this study, we introduce a deep learning-based data-efficient fitness prediction tool to steer protein engineering. Our methodology establishes a lightweight graph neural network scheme for protein structures, which efficiently analyzes the microenvironment of amino acids in wild-type proteins and reconstructs the distribution of the amino acid sequences that are more likely to pass natural selection. This distribution serves as a general guidance for scoring proteins toward arbitrary properties on any order of mutations. Our proposed solution undergoes extensive wet-lab experimental validation spanning diverse physicochemical properties of various proteins, including fluorescence intensity, antigen-antibody affinity, thermostability, and DNA cleavage activity. More than 40% of PO_SCPLOWROTC_SCPLOWLGN-designed single-site mutants outperform their wild-type counterparts across all studied proteins and targeted properties. More importantly, our model can bypass the negative epistatic effect to combine single mutation sites and form deep mutants with up to 7 mutation sites in a single round, whose physicochemical properties are significantly improved. This observation provides compelling evidence of the structure-based models potential to guide deep mutations in protein engineering. Overall, our approach emerges as a versatile tool for protein engineering, benefiting both the computational and bioengineering communities.

bioengineering↗

bioRxiv2023

A New Paradigm for Applying Deep Learning to Protein-Ligand Interaction Prediction

Protein-ligand interaction prediction poses a significant challenge in the field of drug design. Numerous machine learning and deep learning models have been developed to identify the most accurate docking poses of ligands and active compounds against specific targets. However, the current models often suffer from inadequate accuracy and lack practical physical significance in their scoring systems. In this research paper, we introduce IGModel, a novel approach that leverages the geometric information of protein-ligand complexes as input for predicting the root mean square deviation (RMSD) of docking poses and the binding strength (the negative value of the logrithm of binding affinity, pKd) with the same prediction framework. By incorporating the geometric information, IGModel ensures that its scores carry intuitive meaning. The performance of IGModel has been extensively evaluated on various docking power test sets, including the CASF-2016 benchmark, PDBbind-CrossDocked-Core, and DISCO set, consistently achieving state-of-theart accuracies. Furthermore, we assess IGModels generalization ability and robustness by evaluating it on unbiased test sets and sets containing target structures generated by AlphaFold2. The exceptional performance of IGModel on these sets demonstrates its efficacy. Additionally, we visualize the latent space of protein-ligand interactions encoded by IGModel and conduct interpretability analysis, providing valuable insights. This study presents a novel framework for deep learning-based prediction of protein-ligand interactions, contributing to the advancement of this field. O_TEXTBOXKey MessagesO_LIWe introduce the first framework for simultaneously predicting the RMSD of the ligand docking pose and its binding strength to the target. C_LIO_LIIGModel can effectively improve the accuracy of identifying the near-native binding poses of the ligands, and can still outperform most baseline models in scoring power, ranking power and screening power tasks. C_LIO_LIIGModel is still ahead of other state-of-the-art models in the unbiased data set and the target structure predicted by AlphaFold2, proving its excellent generalization ability. C_LIO_LILatent space provided by IGModel learns the physical interactions, thus indicating the robustness of the model. C_LI C_TEXTBOX

bioinformatics↗