1983results about "Molecular structures" patented technology

A system is disclosed for encoding a data string of a first modality into a hierarchical tokenized representation for processing by a text-based deep neural network (DNN) trained on a second modality. The data string comprises multiple units, each having one or more attributes. Each attribute is represented in the tokenized string as a sequence of hierarchical tokens, with a first hierarchical token encoding one or more most significant bits and a subsequent hierarchical token encoding one or more less significant bits. The DNN processes the data string bidirectionally, across the sequence of units and within the token hierarchy, to select tokens that capture attribute information. The selected hierarchical tokens output by the DNN from a representation of the original data string that preserves attribute detail while enabling cross-modal processing using models trained on text.

The invention discloses a colorectal cancer drug relocation method based on multi-omics integration. The system comprises a multi-omics data acquisition and preprocessing module, a tumor microenvironment analysis module, a specific disease network construction module, a multi-dimensional drug relocation module and a result evaluation module. And the tumor microenvironment analysis module comprises cell heterogeneity identification, cell map construction, cell annotation and tumor cell subset annotation. The specific disease network construction module comprises tumor feature expression program extraction, expression program screening, meta-program construction, clinical related meta-program recognition and specific disease protein interaction network construction. And the multi-dimensional drug relocation module comprises a module for identifying diseases by using a random walk algorithm, carrying out drug screening based on disturbance data, carrying out drug screening based on network proximity and carrying out comprehensive drug relocation. From the perspective of single cell data, element programs related to colorectal cancer survival are excavated, corresponding modules are designed, and the efficiency and precision of colorectal cancer targeted drug screening are improved.

The invention belongs to the field of bioinformatics, and relates to an improved integrated deep learning cell communication ligand-receptor interaction prediction method. The method comprises the following steps: firstly, carrying out extraction and dimensionality reduction on biological sequence features of a ligand and a receptor, and constructing multi-modal feature input; secondly, constructing an improved deep neural network branch, introducing a batch normalization layer and a Leaky ReLU activation function, solving the problems of gradient disappearance and neuronal necrosis, and improving regularization strength to prevent overfitting; meanwhile, an enhanced heterogeneous graph auto-encoder branch is constructed, the graph embedding dimension is remarkably expanded to improve the feature capacity, and full convergence of the model is ensured by increasing training rounds; thirdly, fusing the improved deep network with the prediction probability of a heterogeneous graph auto-encoder by adopting a weighted integration strategy; and finally, outputting a potential interaction relationship based on the fusion probability. By optimizing the architecture and the strategy, the prediction accuracy and robustness are remarkably improved, and a reliable tool is provided for analyzing a complex cell communication network.

The invention provides a method for identifying PFAS in an environment based on machine learning pseudo-targeted screening, which comprises the following steps: calling mass spectrum data containing PFAS compounds from a preset mass spectrum database, and performing interference peak elimination processing on the mass spectrum data to obtain a model training data set; extracting a feature data set for model training from the model training data set based on a feature extraction standard; training a plurality of machine learning classification models based on the feature data set, and performing performance evaluation on each machine learning classification model based on a training result to determine an optimal machine learning classification model; and analyzing the optimal machine learning classification model, determining key features when the PFAS is screened and identified, and carrying out PFAS screening identification verification on the optimal machine learning classification model according to the key features based on an actual environment sample. The method has the advantages of saving analysis cost, improving analysis efficiency and improving compound recognition accuracy.

The invention provides a drug-target interaction prediction method and device based on a dynamic heterogeneous double-flow graph neural network, and belongs to the field of drug research and development. The method solves the problems of low prediction accuracy and weak generalization ability caused by insufficient graph structure construction and feature expression in the prior art, and comprises the following steps: constructing a graph structure according to feature data of drugs and targets, and dynamically adjusting weights of the graph structure and edges according to data change to obtain a dynamic heterogeneous graph; the double-flow graph neural network is utilized to process feature information of the medicine and the target spot at the same time, and a complex mode of medicine-target spot interaction is effectively captured; in combination with long-range dependency modeling and random walk feature learning, on the basis of heterogeneous graph convolutional network learning, processing a remote dependency relationship between a drug and a target spot, and capturing multi-hop information in an isomorphic network by using random walk to further optimize interaction prediction; designing a decoder based on matrix completion; training and optimizing the model; the method is applied to drug-target interaction prediction.

The invention discloses a protein generation model optimization method based on deep learning, and relates to the technical field of protein generation. The optimization method comprises the following steps: generating a candidate protein electron density distribution diagram based on the structural characteristics of a target spot by adopting a pre-trained diffusion model; converting the candidate protein electron density distribution diagram into a corresponding amino acid sequence; determining a functional index value corresponding to the amino acid sequence, and constructing a feedback data set; setting a reward threshold value, and marking the feedback data set as a positive sample data set and a negative sample data set according to the reward threshold value; constructing a utility function taking the reward threshold as a reference point, and respectively calculating utility values of the positive sample data set and the negative sample data set; and performing iterative optimization on the diffusion model according to the utility value. By adopting the technology provided by the invention, the dependence on preference paired data of large-scale and high-quality protein sequences can be avoided, and the performance of the protein generation model can be effectively and continuously improved.

The present invention relates to a computer-implemented multi-module classifier method and system for providing a pathogenicity classification score of a variant of a protein of interest. The classifier comprises a sequence module based on a protein language model (PLM); a structure module based on a graph neural network (GNN); a property module; and a unified head module based on a machine learning model. The invention further relates to methods for preparing, training, and implementing the multi-module classifier system.

The invention is applicable to the technical field of clinical medicine, provides a clinical multi-modal cancer drug response prediction method based on feature reconstruction, constructs a clinical multi-modal model for drug response prediction of diffuse large B-cell lymphoma, and aims to predict the drug response of diffuse large B-cell lymphoma by integrating gene sequencing and clinical multi-modal data. And accurate drug reaction prediction is realized. The model adopts an end-to-end multi-stage processing flow: firstly, extracting gene features through TransP-Net, and processing multi-modal clinical data by using a clinical information encoder; then, pseudo-gene features are generated through a clinical-genome filling module to deal with the data missing problem; and finally, multi-modal deep fusion is realized through a clinical information decoder, and a prediction result is output. According to the method, data characteristics and working processes in a real clinical environment are fully considered, two conditions of complete gene data and missing gene data can be processed at the same time, and the method has a good clinical transformation prospect and application value.

The invention relates to the field of artificial intelligence assisted drug discovery, in particular to a protein large language model-based molecule generation and optimization method, which comprises the following steps of: acquiring amino acid sequence information and three-dimensional structure information of a target protein pocket; encoding the amino acid sequence of the protein pocket by using a protein encoder constructed based on a protein large language model to obtain a protein pocket feature vector; using a context encoder module to encode the context information according to a preset molecule generation mode (de novo generation or optimization based on a seed compound) to obtain a latent vector; and fusing the protein pocket feature vector with the latent vector. According to the method, accurate representation of the protein pocket is realized by utilizing the protein large language model, and a generation-screening-optimization iterative drug design strategy is developed by supporting a unified framework of two generation modes, so that the targeting specificity of generated molecules and the overall efficiency of drug design are improved.

The invention discloses a bidirectional reversible conversion method and system between a peptide molecule SMILES and a sequence expression. The core innovation lies in that a new sequence description syntax is defined to retain information of a polypeptide special bond and specific modification of amino acid; a main chain atom index and adjacency traversal topology identification algorithm is adopted, and end group and topology integrated detection and coding are carried out; a residue recognition algorithm for main chain cutting and template library matching is compatible with any standard or non-standard amino acid residues, an extensible end group library / monomer template library and an automatic increment mechanism, and automatic recognition and sequence annotation of S-S disulfide bonds; the invention relates to a high-fidelity assembly algorithm of HELM anchor points and topology aware cyclic peptide processing. The method solves the problems of incapability of supporting a complex polypeptide topological structure, poor reversibility, insufficient expansibility of a monomer library and the like in the prior art, can be widely applied to scenes of quantitative structure-activity relationship model construction, large-scale polypeptide data cleaning and the like, and has remarkable practicability and innovativeness.

To provide a pharmaceutical composition for treating a cancer patient having a tumor having a total passenger gene mutation amount larger than the background mutation amount of the tumor.SOLUTION: A pharmaceutical composition for treating a subject having a tumor with a total passenger gene mutation load that is greater than the background mutation load of the tumor, wherein the background mutation load has been determined based on randomly selected genes of the tumor, comprising antibodies that bind to PD1 as an active ingredient. Antibodies that bind PD1 comprise a heavy chain variable region (HCVR) comprising the amino acid sequence of SEQ ID NO: 21 and / or comprise a light chain variable region (LCVR) comprising the amino acid sequence of SEQ ID NO: 22.SELECTED DRAWING: Figure 1

The present disclosure relates to systems, non-transitory computer-readable media, and methods that utilizing compound-protein machine learning representations to generate target results. For example, the disclosed systems can utilize a compound-protein interaction machine learning model to generate a compound-protein machine learning representation for compound protein pairs. The disclosed systems can utilize the compound-protein machine learning representation to train and utilize other target machine learning models in generating predicted bioactivity results. For example, the disclosed systems train a target machine learning model from compound-protein machine learning representations to generate ADMET predictions and / or biological perturbation program predictions. Furthermore, the disclosed systems can utilize one or more explainability models in conjunction with target machine learning models trained based on compound-protein machine learning representations to identify proteins that contribute to predicted bioactivity results.

The invention provides a parallel multi-target random parameter optimization method and device.The parallel multi-target random parameter optimization method comprises the steps that the number of nodes of a parallel computing cluster is configured, and a random seed sequence is generated; carrying out structure loading and compiling on the to-be-optimized model, and obtaining global configuration of the to-be-optimized model; sending the global configuration to all processes, generating an initial parameter population meeting the constraint of the global configuration by each process based on the random seed, and performing serial evaluation to obtain a target function value; independently iteratively executing shuffling complex evolution by each process until a convergence condition is met; each process outputs the respective optimal parameter vector and the corresponding objective function value to a root process, and a global optimal solution is determined through the root process. By means of the method and device, the accuracy and efficiency of parameter optimization can be remarkably improved, and the problem that the model optimization effect is poor in the prior art is solved.

Systems and methods for characterizing an interaction between a compound and a polymer include obtaining a plurality of sets of atomic coordinates. Each set of atomic coordinates comprises the compound bound to the polymer in a corresponding pose in a plurality of poses. Each respective set of atomic coordinates, or an encoding thereof, is sequentially inputted into a neural network, to obtain a corresponding initial embedding as output, thereby obtaining a plurality of initial embeddings. Each initial embedding corresponds to a set of atomic coordinates in the plurality of sets of atomic coordinates. An attention mechanism is applied to the plurality of initial embeddings, in concatenated form, to obtain an attention embedding. A pooling function is applied to the attention embedding to derive a pooled embedding. The pooled embedding is inputted into a model to obtain an interaction score of the interaction between the compound and the polymer.

A method for judging the structural stability of a disease marker belongs to the field of disease markers, and comprises the following steps: acquiring a disease marker sample, and collecting spectral data, fluorescence intensity data, surface charge data and protein content data of the disease marker as multi-dimensional data of the disease marker; performing spectral stability analysis on the multi-dimensional data of the disease marker to obtain a first judgment result, and if the first judgment result is in a stable state, performing molecular structure scoring analysis on the fluorescence intensity data of the disease marker through a pre-trained deep neural network model to obtain a second judgment result; and if the second judgment result is a stable state, performing structural integrity analysis based on the surface charge data of the disease marker and the protein content data of the disease marker to obtain a third judgment result, and outputting a final structural stability judgment result of the disease marker according to the third judgment result. The technical problem that the overall structure stability of the disease marker is difficult to accurately assess in the prior art can be solved.

The invention discloses an administration agent in-vivo release and absorption visualization method based on a CFD-PBM coupling model. The method comprises the following steps: constructing a CFD model of a physiological environment of an injection site; establishing a population balance model (PBM) of the drug particles; the PBM is embedded into a CFD model solver, multi-scale coupling simulation is carried out, and a CFD-PBM coupling model is obtained; the three-dimensional visualization engine dynamically displays drug concentration distribution and particle behaviors; reversely adjusting preparation prescription parameters by using a multi-parameter optimization algorithm; and calibrating model parameters, and generating a key data report. According to the method, the prediction precision and the research and development efficiency can be remarkably improved, and the method is suitable for development of long-acting preparations such as microspheres and implants.

The invention discloses a molecular simulation method and device, a nonvolatile storage medium and electronic equipment. The method comprises the following steps: acquiring molecular system information of a current molecular configuration; calculating the first energy and the first gradient of the key atom region by adopting a variable component sub-feature solicitation solver algorithm, and calculating the second energy and the second gradient of the non-key atom region; wherein the first gradient is used for expressing the stress condition of each atom in the key atom area; the second gradient is used for expressing the stress condition of each atom in the non-key atom area; and updating the current molecular configuration according to the first energy, the first gradient, the second energy and the second gradient to obtain a target molecular configuration. According to the method and the device, the technical problem that the accuracy of the simulation result cannot be ensured under the condition that the computing resources are limited due to excessive computing resources consumed by the quantum chemical computing mode when the molecules are simulated by adopting the quantum chemical computing mode in the related technology is solved.

The invention discloses a drug-target interaction prediction method based on double-flow collaborative attention and sparse feature fusion, and belongs to the technical field of computational biology. The method solves the problem that the existing method cannot capture the sub-structure discrimination features and the key binding region features of the drug-target interaction pair. According to the method, a double-flow collaborative attention strategy combining a multi-scale space attention mechanism and a channel enhanced attention mechanism is adopted to cooperatively capture discriminative features of substructures, the multi-scale space attention mechanism utilizes a multi-branch convolutional layer to adaptively integrate space substructure features, and molecular representation of each substructure is enhanced; the channel-enhanced attention mechanism mitigates the inconsistency of substructure features. The sparse attention mechanism can highlight the key features while suppressing the noise, and the cross attention mechanism improves the extraction capability of the features of the key combination region through feature interaction between the modeling drug and the target. The method can be applied to drug-target interaction prediction.

The invention provides an anti-cancer active component optimization method for breast cancer treatment, which comprises the following steps: analyzing a potential interference path of an anti-cancer active component on immune system cell viability by adopting a computational chemistry simulation method according to a preliminary structural function mapping relationship, and determining a specific molecular mechanism range of immune system weakening; according to the synergistic effect evaluation result, optimizing the structural parameters of the active components through a molecular docking algorithm, adjusting the binding affinity of the active components with tumor cell targets and pathogenic bacteria targets, and determining a final structural optimization scheme; aiming at the final structure optimization scheme, verifying the expression of the modified active component on tumor inhibition and antibacterial ability by adopting simulation data of an in-vitro activity test, and obtaining a verification data set of comprehensive performance; and aiming at the updated active component design data, through a multi-objective optimization model, balancing the synergism of an anti-cancer effect and a health protection mechanism, and determining a final compound structure configuration suitable for complex requirements of a clinical environment.

The invention relates to the technical field of medicine data analysis, in particular to a method, a system and equipment for generating a medicine relocation research report and a medium. According to the method for generating the drug relocation research report, a drug relocation request submitted by a pharmaceutical enterprise is decomposed into three sub-tasks of indication extension analysis, target matching analysis and drug effect simulation verification, and an indication extension analysis agent, a target matching analysis agent and a drug effect simulation verification agent are called respectively; the contribution value of each link is calculated through multi-agent cooperation, and finally a relocation research report containing potential rating and confidence evaluation is generated based on the drug-disease knowledge graph. By implementing the technical scheme provided by the invention, the efficiency and reliability of drug relocation analysis are improved, and scientific and credible decision support is provided for pharmaceutical enterprises.

The invention relates to application of mitogen-activated protein kinase PiPmk1 in screening molecules for inhibiting formation of a phytophthora infestans infection structure. The invention finds that the molecular marker plays a key role in the development of a phytophthora infestans cyst bud tube and / or the formation of an infection structure, which indicates that the molecular marker has the potential of serving as a target for inhibiting phytophthora infestans infection, and a corresponding molecular screening method and a generative prediction model are further developed and utilized; rapid and accurate prediction and screening of phytophthora infestans infection inhibitory molecules are realized, the research and development period and cost are greatly shortened, and a new thought and a new method are provided for development of accurate targeted green pesticides.

The invention discloses a multi-modal data fusion drug-target affinity prediction system based on Graph Transform. The system comprises a data preprocessing module, a graph representation module, a text representation module and an affinity prediction module. The data preprocessing module is responsible for analyzing and processing the SMILES character string of the compound and the protein ID, and extracting the structural information of the compound and the three-dimensional structural data of the protein. And the affinity prediction module performs multi-modal fusion on the graph features and the text features, processes fusion feature vectors through a feedforward neural network comprising three full-connection layers, and outputs a prediction result of drug-target affinity. According to the method, multi-modal information of a graph structure and a sequence text is combined, the accuracy of cross-domain drug-target affinity prediction can be effectively improved, and the method has a wide application prospect.

The invention provides an antibody drug conjugate property prediction method based on multi-modal fusion, and belongs to the field of bioinformatics. The method comprises the following steps: firstly, explicitly modeling sequence position information through sine position coding; secondly, introducing a bidirectional cross attention mechanism to establish interaction between a light chain and a heavy chain and alignment between an antigen and an antibody; thirdly, the integrated graph neural network reconstructs the adjacency relation according to the attention weight, and topological features are extracted; and finally, in combination with a double-stage self-adaptive refining module, two-stage treatment of alignment and refining is realized, and each modal feature contribution is adjusted in a self-adaptive manner. And meanwhile, the sequence robustness is improved through Mask perception feature extraction, and the interpretability analysis of the key binding sites is realized through the attention weight. The method can significantly improve the prediction accuracy, and can be widely applied to cancer targeted therapy and drug design optimization.

The invention discloses a computer-aided drug screening method, system or device based on FBXO2 and PKM2. The invention provides a brand-new method, system or equipment for screening oral squamous cell carcinoma treatment drugs based on FBXO2 and PKM2, provides a tool for new drug development and clinical application for treatment of oral squamous cell carcinoma, and has a wide application prospect.

The invention belongs to the field of bioinformatics, and relates to a protein post-translational modification prediction method based on multi-modal deep learning. The method comprises the following steps: firstly, performing multi-modal feature extraction by inputting a protein sequence and three-dimensional structure data to obtain a sequence feature vector and a structure feature vector; secondly, carrying out feature fusion by adopting a cross-modal attention mechanism and a self-adaptive gating network; then, combining the fusion features with the disease type information, and performing fine adjustment on the prediction probability through a disease specific coding network; then, using a multi-task learning framework to predict the locus probabilities of various protein post-translational modification types in parallel; finally, feature importance is calculated through a gradient back propagation technology, and a comprehensive report is output in combination with variation influence analysis. According to the method, high-precision and explainable protein post-translational modification prediction with disease perception capability is realized, and an important calculation and analysis tool is provided for revealing a disease molecular mechanism and finding accurate drug targets.

The invention belongs to the field of medical technology analysis, and discloses an analysis method and system for revealing a hidden binding pocket of a drug target, and the method comprises the steps: firstly obtaining a representative conformation metastable state of a target protein through conventional molecular dynamics simulation and clustering analysis; secondly, constructing a Markov state model to analyze a dynamic transformation rule between conformations; carrying out enhanced sampling by adopting meta-dynamics, and deeply exploring a rare conformation space; and finally, constructing a free energy landscape to quantitatively evaluate the relative stability of the conformation, and identifying a hidden binding pocket in the stable rare conformation. According to the method, the limitation of a single calculation means is overcome, a full-chain calculation system of dynamic conformation analysis-hidden cavity feature mining-novel ligand rational design is constructed, and the formation mechanism and potential druggability of the hidden pocket can be comprehensively revealed from the two dimensions of dynamics and thermodynamics; and an efficient and accurate calculation framework is provided for research and development of innovative drugs targeting difficult drug targets.

The invention relates to the field of gene engineering. In particular, the present invention relates to cytosine deaminases and their use in base editing. More specifically, the invention relates to a method for screening and identifying a deaminase, a base editing system based on a newly identified cytosine deaminase, a method for editing a target sequence in a genome of an organism (such as a plant) by using the base editing system, and a method for screening and identifying the target sequence. As well as genetically modified organisms (e.g., plants) and progeny thereof produced by the method.

Disclosed herein are predictive models for T-cell epitope prediction, B-cell epitope prediction, and protein design wherein a method is implemented for generating a protein variant amino acid sequence of a target protein having one or more modified properties, the method comprising: (a) iteratively sampling an input amino acid sequence of the target protein, and (b) sampling the individual protein score of at least one weighted relative contribution of the single residue mutant input amino acid sequence to the at least one target property across a plurality of other single residue mutant input amino acid sequences to generate a combined protein score, wherein the combined protein score corresponds to the protein variant comprising one or more amino acid mutations of the single residue mutant input amino acid sequences.

The invention relates to the technical field of drug safety evaluation, in particular to a multi-center combined evaluation method and system for drug safety based on a graph neural network and federated learning. Known and unknown drug interaction is systematically predicted based on a drug multi-relation knowledge graph and a graph neural network, a key path of DDI is identified through a graph attention mechanism, a molecular mechanism of the interaction is revealed, and natural language interpretation is generated. And meanwhile, through privacy protection and multi-center cooperation, a federal learning architecture is utilized to break data islands and improve the external effectiveness of an evaluation conclusion on the premise of protecting patient privacy and meeting data compliance requirements. The heterogeneity of data of different mechanisms is effectively evaluated through distribution deviation detection, deviation caused by blind extrapolation is avoided, a real-world evidence methodology report and a data traceability auditing clue are automatically generated, the requirement of a supervision mechanism for real-world evidence quality is met, medicine supervision decision is supported, and medicine research, development and review are accelerated.

The invention relates to a UGT enzyme subtype substrate specificity prediction method and model. On the basis of a directional message passing neural network, graph structure characterization of a small molecule compound and features of specific protein binding sites of UGT enzyme are deeply fused, a bimodal prediction normal form of'molecule + protein binding sites' is designed, a deep learning model is constructed, conversion from compound center prediction to molecule-enzyme binding site comprehensive prediction is achieved, and the prediction accuracy is improved. And accurate classification prediction can be carried out on UGT enzyme substrates and non-substrates.