30 results about "Functional prediction" patented technology

The invention relates to the technical field of artificial intelligence, and provides a protein function prediction method and device based on multi-modal protein data, and the method comprises the steps: obtaining protein multi-source data, carrying out the feature extraction of a protein sequence in the protein multi-source data, and obtaining a protein sequence feature; constructing a heterogeneous graph based on the protein multi-source data; performing feature coding on the heterogeneous graph by adopting a graph attention mechanism to obtain protein graph features; performing multi-modal fusion on the protein sequence features and the protein map features by adopting a gating fusion mechanism to obtain fusion features; and performing multi-label classification prediction based on the fusion features to obtain a protein function annotation result. The accuracy and robustness of protein function prediction can be improved, and the problems that in the prior art, multi-source protein data cannot be effectively integrated, and the method is sensitive to data noise are solved.

The invention discloses a new mutation-oriented pathogenicity prediction method, which comprises the following steps of: acquiring new mutation site data to be analyzed, and generating a numerical feature vector containing evolution conservative property, function prediction score and quantitative clinical evidence; constructing a feature extractor through the numerical feature vector; inputting a training data set containing a label sample and a label-free sample into the feature extractor, and updating the network parameters of the feature extractor and each header; and predicting target variation by using the feature extractor after parameter updating and the supervised classification head to obtain a pathogenicity probability, and calculating a contribution value of each feature in the numerical feature vector to the pathogenicity probability. According to the technical scheme provided by the invention, the auxiliary decision-making information conforming to the diagnosis habit of a doctor is output, the workload of manually interpreting the variation with unclear meaning is greatly reduced, and the end-to-end intelligent processing from the original sequencing data to the clinical auxiliary diagnosis is realized.

The invention belongs to the crossing field of marine biotechnology and bioinformatics, and particularly relates to a method for high-throughput identification of natural functional oligopeptides. At present, bottlenecks still exist in development and application of various functional oligopeptides such as penetrating peptides and antibacterial peptides in non-model species, and the high adaptability of the existing functional oligopeptides in complex and diverse non-model organisms is limited mainly due to differences (amino acid preference and receptor specificity) among species. Therefore, the invention provides a method for high-throughput identification of functional oligopeptides from natural sequences of organisms by means of machine learning. According to the method, based on a biological natural protein sequence, an oligopeptide sequence set meeting a preset length condition is generated through a sliding window strategy system, a deep machine learning model is utilized to perform high-throughput functional prediction on the oligopeptide sequence, and on the basis of performing'interruption-prediction-re-comparison 'on existing functional oligopeptides, the functional oligopeptide sequence set meeting the preset length condition is obtained. The comparison threshold value for accurately identifying the functional oligopeptide can be determined, and the natural functional oligopeptide existing in the protein sequence can be effectively identified based on the threshold value. The functional oligopeptide obtained by the method has strict screening of physicochemical properties of a machine learning model and high-adaptability biological characteristics formed in species evolution. According to the method, the direct application capability of the functional oligopeptide in non-model species is remarkably improved, and an accurate and high-throughput technical means is provided for efficient development and application of the functional oligopeptide in non-model organisms.

The invention discloses a microbial data semantic processing method and system based on a graph model, and relates to the technical field of knowledge graph and microbial information processing, and the method comprises the steps: obtaining microbial data from a heterogeneous data source, and carrying out the standardization processing; constructing a domain ontology model conforming to microbial taxonomy specifications; implementing entity disambiguation and unified identifier mapping by utilizing a pre-training language model; a semantic relation triple is extracted in a mode of combining remote supervision and deep learning; constructing a microorganism knowledge attribute graph and storing the graph in a graph database; knowledge reasoning and function prediction are carried out by using a graph attention network; the method supports natural language semantic retrieval, and solves the problems that heterogeneous microorganism data integration is difficult, the naming disambiguation precision is insufficient and the function association mining capacity is limited.

The application provides a method and system for generating and predicting the function of mRNA untranslated region sequence based on coding sequence. The method comprises: constructing a pre-training data set and a downstream task data set; using the pre-training data set to perform autoregressive training on the constructed conditional generation model to obtain a UTR sequence generation model; using the downstream task data set to fine-tune the UTR sequence generation model to obtain a downstream task function prediction model; generating candidate UTR sequences based on the UTR sequence generation model, and evaluating the generation ability of the UTR sequence generation model in the non-coding region sequence generation task; and predicting the function attribute of the UTR sequence based on the downstream task function prediction model, and evaluating the prediction ability of the downstream task function prediction model in multiple UTR related downstream tasks. The application considers the synergistic relationship between UTR and CDS in the UTR sequence generation process, and improves the efficiency and rationality of UTR sequence design.

The application provides an RNA sequence classification method based on a Mamba model and semi-supervised learning, and belongs to the field of bioinformatics. First, multi-scale sparse features are extracted from the RNA sequence, high-dimensional features are compressed into low-dimensional potential space vectors through an encoder network, and L2 normalization is performed to enhance feature separability. Second, a Mamba model based on a selective state space model is introduced in a residual structure to efficiently model long-range dependencies with low complexity. Finally, a semi-supervised learning of unsupervised feature reconstruction and supervised classification is realized by constructing an encoder-decoder structure, and a weighted loss function is used to improve the generalization performance of the model under small sample conditions. Experimental results show that the method significantly improves the F1 score in the multi-class RNA sequence classification task. The application reduces the cost of long sequence modeling and labeling, and is suitable for the fields of RNA function prediction, biomedical research and disease diagnosis.

The present disclosure provides a rational modification method of spermidine synthase (SpeE) in Bacillus subtilis, which relates to the technical fields of enzyme engineering and computational structural biology. The modification method comprises the following steps: constructing a high-confidence three-dimensional model of Bacillus subtilis SpeE, determining the key action scope by double-substrate (S-adenosyl methionine and putrescine) docking, performing virtual saturation mutation on the residues in the region, screening the optimal mutant by combining function prediction, stability analysis, molecular dynamics simulation and binding free energy calculation, and explaining the mutation mechanism from the atomic level. The SpeE mutant (such as G85S, G85A) obtained by the modification method has significantly improved catalytic efficiency and substrate binding affinity, and when applied to a whole-cell catalytic system, it can greatly improve the yield of spermidine, and is suitable for the industrialized green biosynthesis of spermidine.

The invention relates to the field of bioengineering and genetic engineering, and particularly provides a rapid digging method of isomerase genes. According to the method, on the basis of big data analysis and bioinformatics technologies, in combination with means of sequence alignment, gene expression analysis, function prediction and the like, rapid and accurate mining of isomerase genes is achieved, and powerful technical support is provided for development and application of the enzyme engineering field.

The invention relates to the technical field of gene analysis, and discloses an intelligent cancer-related gene analysis method based on weighted voting. According to the method, firstly, different calculation biological tools are used for performing preliminary prediction on a to-be-detected gene, then prediction scores of the multiple calculation biological tools are fused in a non-linear feature construction mode, and finally the prediction scores, confidence coefficients and non-linear extension features of to-be-detected variation points are input into a plurality of pre-trained first-level models. And obtaining output results of the plurality of primary models. And fusing output results of different primary models by adopting a weighted fusion method, and analyzing to obtain a function prediction result corresponding to the to-be-tested gene data based on a final fusion result. Therefore, the to-be-detected gene data can be analyzed more comprehensively, analysis results of different analysis tools and different models are integrated, and finally the reliability of the analysis results is improved on the whole.

This invention relates to a method for targeted design and optimization of the 5′ untranslated region (UTR) of mRNA based on generative language models and reinforcement learning. The method includes: constructing pre-training data for the 5′ UTR; obtaining a prediction task dataset; calculating the minimum free energy (MFE) corresponding to the pre-training data; constructing a generative language model and performing phased pre-training on the generative language model using the pre-training data and the MFE to obtain a generative model and a prediction task fine-tuning model; fine-tuning the prediction task fine-tuning model based on the prediction task dataset to obtain a functional prediction model; evaluating the generative model's generation capability in the 5′ UTR sequence generation task; and performing targeted optimization design of the 5′ UTR sequence based on a reinforcement learning framework, combining the generative model and the functional prediction model. This invention can efficiently generate 5′ UTR sequences with specific biological functions, providing strong support for functional mRNA design and showing broad application prospects.

The application discloses a kind of polypeptide function prediction method and device.The method includes data set construction and data preprocessing;Establish multi-label prediction model, by feature embedding module, multi-scale convolutional neural network CNN module, BiGRU module for processing context-related sequence and classification module are composed;The network model constructed is trained using the obtained data set, and the optimization algorithm is used to improve the model feature extraction and parameter prediction performance;Then the performance of the multi-label prediction model is evaluated by calculating the accuracy, coverage, accuracy, absolute true value and absolute false value five indexes.The application effectively utilizes the correlation between labels, improves the accuracy of polypeptide function prediction, the method is strong in operability, strong in practicality, can be applied to the function prediction of eight kinds of bioactive peptides simultaneously, is a reasonable and effective prediction method, the precision of eight polypeptide functions predicted by the application is as high as 80.4%.

The invention discloses an identification method of a sequence participating in potato centromere relocation, and belongs to the technical field of plant heredity and molecular biology. According to the method, T2T-level genome data and CENH3 ChIP-seq data of different haplotypes of potatoes are integrated, sequence screening, annotation and function prediction are carried out by utilizing a bioinformatics analysis tool, and a satellite DNA sequence which is 2-3Kbp in length, is highly conserved in the same haplotype and can be specifically combined with CENH3 protein is identified. The sequence can induce CENH3 to deposit in a non-native centromere area, so that functional relocation of centromere is realized, and a core molecular element and a technical support are provided for artificial centromere construction, plant chromosome engineering and accurate genome operation. The method solves the problems that in the prior art, a functional sequence for clearly inducing centromere relocation is lacked, a CENH3 targeted recruitment mechanism model is not established and the like, and has a wide application prospect.

The invention aims to provide a method, equipment, medium and program product for determining a target enzyme expressed in a target host, and the method comprises the following steps: on the basis of amino acid sequence information and structure information corresponding to a heterologous enzyme meeting a required function and information of the target host, predicting the target enzyme by using a codon sequence prediction model, the codon sequence prediction model is used for determining codon sequence information and corresponding score information, the score information is used for comparing expression quantities of different codon sequences, the enzyme corresponding to the codon sequence with the higher score can have the higher expression quantity in the target host, and the enzyme corresponding to the codon sequence with the higher score can have the higher expression quantity in the target host. Furthermore, the enzyme efficiently expressed in the target host can be quickly screened out, and the screening cost is reduced. In addition, an enzyme function prediction model and / or an enzyme-substrate interaction prediction model can be combined to screen from the aspects of expression quantity, catalytic reaction type and / or enzyme activity and the like, so as to obtain the enzyme which is high in expression and activity and has required functions.

The application discloses a kind of predictable phaseolus vulgaris response melatonin regulation SSR molecular markers and its application, it is related to biotechnology field.The application carries out the effectiveness identification of ordinary phaseolus vulgaris natural germplasm resources to melatonin treatment, the efficiency of melatonin treatment-cell wall-salt tolerance in ordinary phaseolus vulgaris germplasm resources is analyzed, and using the reference genome in Esembl plants database, a complete set of SSR molecular markers is developed, and according to transcriptome information, the SSR marker near response gene is mined, through polymorphism analysis and single marker analysis, two SSR markers with significant difference associated with melatonin traits are screened, can be used as the molecular marker for predicting phaseolus vulgaris response melatonin regulation, and provides technical support for the function prediction of melatonin treatment in the sprouting stage of ordinary phaseolus vulgaris.

The application discloses a pathogenicity prediction method for novel variation, comprising the following steps: obtaining novel variation site data to be analyzed, and generating a numerical feature vector containing evolutionary conservation, functional prediction score and quantitative clinical evidence; constructing a feature extractor through the numerical feature vector; inputting a training data set containing labeled samples and unlabeled samples into the feature extractor, updating the feature extractor and the network parameters of each head; applying the feature extractor and the supervised classification head after parameter updating to predict target variation, obtaining pathogenicity probability, and calculating the contribution value of each feature in the numerical feature vector to the pathogenicity probability. The technical scheme provided by the application outputs auxiliary decision information conforming to the diagnosis habit of doctors, greatly reduces the work load of artificial interpretation of unknown variation, and realizes end-to-end intelligent processing from raw sequencing data to clinical auxiliary diagnosis.

The present application relates to a kind of fusion sequence and structural multimodal feature and the classification method of antihypertensive peptide combined with contrast-generative joint optimization, comprising: the sequence characteristic representation and structural characteristic representation of peptide segment are extracted;The sequence characteristic representation and structural characteristic representation are contrast-generative joint optimization multimodal feature enhancement;Using preset Kan-Conv structure and label smoothing joint optimization classification model based on, the enhanced feature representation is classified into peptide.This application realizes the high-precision identification of antihypertensive peptide functional activity, this method innovatively utilizes three aspects of information, such as sequence, structure and generative latent space, fills the blank of ignoring structure-sequence synergistic effect in the existing peptide function prediction field, significantly improves the screening efficiency and prediction reliability of antihypertensive peptide.

The invention provides a preparation method and application of a dental pulp stem cell preparation based on AI activity prediction, and the method comprises the following steps: collecting historical preparation parameters and corresponding activity indexes of the dental pulp stem cell preparation, constructing a historical database, and carrying out data preprocessing and core parameter screening; according to the method, an activity prediction model comprising a first sub-model and a second sub-model is constructed, bidirectional function prediction is realized through two-stage training, multiple types of activity indexes and comprehensive scores can be predicted based on preparation parameters, and the preparation parameters can be reversely recommended according to activity requirements; and preparing a compound preparation which takes the dental pulp stem cells as a core and PDD-ALN as a synergistic enhancement carrier based on model output. The method can accurately guide the preparation of the dental pulp stem cell preparation, and is especially suitable for the development of periodontitis treatment drugs.

The invention relates to a small molecule and G protein coupled receptor binding biological activity and function prediction method, and provides a small molecule and G protein coupled receptor binding biological activity and function prediction method based on a deep neural network model and a contrast learning strategy aiming at the limitation of the existing small molecule and protein interaction prediction method. Through fusion of a pre-trained protein language model and a pre-trained molecular representation learning model, key conservative residue interaction mode change is predicted in combination with multi-scale characteristics of pockets and molecular structures and a structure change module, and prediction of small molecule activity and functions is realized. The method has good interpretability, and lays an important foundation for drug virtual screening and artificial intelligence drug design of the G protein coupled receptor.

The present application relates to the field of bioengineering and genetic engineering, and particularly provides a rapid mining method of isomerase genes. The method is based on big data analysis and bioinformatics technology, combined with sequence alignment, gene expression analysis and function prediction, etc. means, realizes the rapid and accurate mining of isomerase genes, and provides strong technical support for the development and application of enzyme engineering field.

This invention belongs to the field of plant rhizosphere research and discloses a method for determining the regulatory effect of exogenous selenium application on the rhizosphere microecology of Amorphophallus bulbifera. The method includes: experimental design and sample collection; soil microbial DNA extraction and amplicon sequencing; and quality control and analysis of the microbiome data. This invention utilizes Illumina MiSeq high-throughput sequencing technology to analyze the changes in the composition and diversity of rhizosphere soil bacteria communities in Amorphophallus bulbifera under exogenous selenium treatment, and further analyzes the functions of related communities through PICRUSt functional prediction analysis. The conclusions are as follows: exogenous selenium treatment induces significant changes in the rhizosphere soil bacterial community of Amorphophallus bulbifera, which is beneficial for creating a stable and more diverse rhizosphere microbial community structure.

The invention provides a protein function prediction method and system, computer equipment and a storage medium, and belongs to the technical field of bioinformatics and new drug research and development, and the method comprises the following steps: obtaining a to-be-predicted protein sequence; traversing each amino acid in the protein sequence, respectively mapping the hydrophobicity, polarity and molecular weight of the amino acid into pitch, tone and duration of music elements, and generating a music score sequence corresponding to each amino acid; performing coding synthesis on the music score sequence to obtain a Mel spectrogram; inputting the Mel spectrogram into a pre-trained double-branch attention fusion model to obtain a visual feature vector and an acoustics context vector; splicing the visual feature vector and the acoustics context vector to obtain a fused feature vector; and obtaining a protein function prediction result according to the fusion feature vector. Through fusion of visual features and acoustic context vectors, complementation of structural information and long-range association is realized, integrity and accuracy of function prediction are improved, and prediction efficiency is improved while data dependence is reduced.

The application provides bone reconstruction regulating polypeptides and application. On the basis of the first experiment of the inventors, bone reconstruction regulating polypeptides are designed and newly synthesized by using solid-phase polypeptide synthesis method through sequence alignment, structure analysis, physicochemical property and function prediction. The polypeptides have the characteristics of short length, simple synthesis, low cost, low cytotoxicity, high biological stability, moderate half-life, good bone targeting, and strong bone resorption regulating ability. The differentiation and function of osteoclasts and osteoblasts can be controlled by adjusting the concentration of the synthesized bone reconstruction regulating polypeptides, the ordered regulation of bone resorption and bone formation is realized, and the polypeptides have a broad potential application prospect.

The invention provides an anti-hypertension peptide function prediction method and system and a medium, and the method comprises the steps: collecting explicit feature data and implicit sequence feature data of a target anti-hypertension peptide, the implicit sequence feature data being obtained through extraction of a protein language model; performing fusion processing on the explicit feature data and the implicit sequence feature data to obtain multi-modal fusion features; taking the obtained binding energy data of the target antihypertensive peptide and the angiotensin converting enzyme as a supervision label, constructing a machine learning prediction model, and performing training; and inputting the multi-modal fusion features into a trained machine learning prediction model to obtain a prediction result of the binding activity of the target antihypertensive peptide and the angiotensin converting enzyme. According to the method, the reliability of the supervision label is optimized through molecular docking and dynamic simulation, and the information dimension is complemented by fusing the ESM2 embedding characteristic and the physicochemical characteristic, so that the function prediction precision and generalization ability of the anti-hypertension peptide are greatly improved, the efficient screening of the high-activity peptide is realized, and the research and development cost is reduced.

The application relates to the technical field of quantum computing, in particular to a protein function prediction method based on a hybrid quantum-classical neural network, which comprises the following steps: obtaining a target protein sequence, extracting a high-dimensional feature vector of the target protein sequence by using a pre-trained classical protein language model and performing dimension reduction processing to obtain a low-dimensional input feature; inputting the low-dimensional input feature into a variational quantum circuit, performing adaptive quantum state coding by using frequency parameters and phase parameters that are optimized through training, and obtaining an initial quantum state; performing multi-scale feature entanglement on the initial quantum state in the variational quantum circuit to obtain an evolved quantum state; performing measurement on the evolved quantum state to obtain quantum feature expectation values, inputting the quantum feature expectation values into a classical classification network, and outputting a function prediction result of the target protein sequence. The scheme can improve the separability of protein function categories in a quantum feature space and reduce the parameter size of the classical classification network.

The invention discloses a plant flavonoid pathway mining and database construction method and system based on a protein language model and multiple omics. The method comprises the following steps: acquiring genomes, proteomes and annotation data of plants of multiple species, constructing a standardized flavonoid resource library, extracting high-dimensional sequence features by utilizing an ESM-2 protein language model, performing accurate flavonoid enzyme prediction in combination with three-dimensional structure modeling and molecular docking, and constructing an extended flavonoid metabolism network map. And a database platform is integrated based on a front-end and back-end separation architecture. According to the invention, the accuracy and coverage range of flavonoid related gene identification are obviously improved; compared with a traditional annotation method based on sequence similarity comparison, the method has better functional distinguishing capacity under the low homologous background, and generalization annotation can be reduced; complementing modification reaction nodes deleted by the metabolic pathway, and enhancing the integrity of pathway annotation; total factor integration of multi-omics data and a metabolic network is realized, and interactive pathway visualization and function prediction are supported.

The invention discloses a cloud analysis process for automatically processing 16S or 18S or ITS data. The cloud analysis process for automatically processing the 16S or 18S or ITS data comprises the steps that the task delivery module achieves webpage triggering submission of tasks, upload files are prepared, the upload files are received for ASV classification, ASV file data analysis is conducted, alpha and beta diversity analysis is conducted, and function prediction analysis is conducted on the upload data. According to the cloud analysis process for automatically processing the 16S or 18S or ITS data, time and effort are saved, manual errors are reduced and presented in various visual modes, a user is helped to better understand the data, a comprehensive bioinformatics analysis solution is provided, and the cloud analysis process has important significance in the fields of research, medical treatment, environmental monitoring and the like.

The present application relates to the technical field of molecular diagnosis, and specifically relates to application of female uterine cavity flora as a marker in endometriosis diagnosis. The present application collects uterine cavity flora of patients, extracts genomic DNA, performs bacterial 16s-rRNA sequencing, uses KEGG as a reference database for function prediction, determines that abnormal abundance of uterine cavity flora dominant bacteria has significant correlation with dysmenorrhea and high CA125 value of endometriosis patients, clarifies the correlation between abnormal uterine cavity flora and endometriosis, and provides a new method for evaluating the risk of endometriosis by detecting uterine cavity flora atlas.

Method for reducing environmental models for the simulation of driving functions using a normalized influence measure, with: - Acquisition of input data consisting of an underlying environment model, a driving function to be tested and a simulation target as well as simulation parameters including initial conditions and time intervals, wherein the input data is fed to an input of a system that performs the subsequent steps; - Providing a predictive model in the form of a neural network trained on historical simulation results to predict a simulation output depending on the environment model and the driving function, wherein the predictive model is parameterized so that it predicts the simulation output exclusively from the input data; - Generating a predicted simulation output; - Applying an explainability metric to the predictive model for the specific prediction in order to determine an influence value for each feature of the environmental model and to scale this to a normalized SHAP value between 0 and 1; - Deriving an adaptation of the environment model from the normalized SHAP values ​​by reducing the resolution for features below a first threshold and maintaining full detail for features above a second threshold, and setting a level of detail proportional to the normalized SHAP value in between, thereby providing an adapted environment model; - Generating a justification and influence output that lists the features relevant to the predicted simulation output and their influence, for documentation and traceability of the model fitting.

The invention discloses an improved Transformer-based carbon cycle related protein function prediction method, which comprises the following steps of: obtaining a to-be-detected carbon cycle related protein amino acid sequence and pretreating to generate a standard to-be-detected amino acid sequence; semantic embedding characteristics of the amino acid sequence to be detected are extracted, and physicochemical property characteristics are calculated; fusing the semantic embedding features and the physicochemical property features to generate comprehensive feature representation; inputting the comprehensive feature representation into an improved Transform network to generate a fixed-length feature vector for capturing global key features; and performing distribution calibration on the feature vectors by using a Gaussian mixture model and variational Bayesian estimation, finally inputting the feature vectors into a full-connection neural network, and accurately predicting protein functions as one of five function categories through a Softmax activation function: carbon fixation, carbon release, organic synthesis, organic degradation and organic conversion. According to the invention, high-precision function prediction can be realized.

The invention discloses an IncRNA function prediction method based on gene ontology structure semantic perception, and relates to an IncRNA function prediction method. In order to solve the problems that according to an existing lncRNA function prediction method, fusion of multi-source heterogeneous information is insufficient, a DAG structure of a GO body is difficult to use at the same time, and a prediction result violates a True Path Rule, firstly, a VAE-based method is used for effectively fusing multi-source heterogeneous features of lncRNA; the method comprises the following steps of: firstly, extracting semantic features of lncRNA by using a BioBERT method, finally, injecting lncRNA information into GO features through cross attention, adopting DAG-LSTM to bidirectionally spread the features, and finally, enabling GO to pay more attention to True Path Rule information by using a self-attention mechanism causal mask. According to the invention, the accuracy of lncRNA function prediction can be effectively improved.