Microbial synthesis nano material recommendation method based on knowledge graph

Abstract

The invention relates to the technical field of bioinformatics and artificial intelligence, in particular to a microbiological synthesis nano material recommendation method based on a knowledge graph. Acquiring microorganism information and nano material information; constructing a knowledge graph through the microorganism information and the nano material information; wherein nodes of the knowledge graph are composed of microorganisms, a nano material, a synthesis method and elements; obtaining a structural feature vector and a semantic feature vector based on the knowledge graph; splicing the structural feature vector and the semantic feature vector to obtain a representation vector of each node; and scoring each node based on the representation vector, and judging the relationship between the microorganisms and the nano material according to the scoring result. According to the method, the recessive association between the microorganism and the potential nano material can be mined from the large-scale knowledge graph.

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics and artificial intelligence technology, and in particular to a method for recommending nanomaterials synthesized by microorganisms based on a knowledge graph. Background Art

[0002] The rapid development of synthetic biology has promoted the widespread application of microorganisms in industrial production. Microorganisms are widely used in the synthesis of nanomaterials. Due to their advantages such as high production efficiency, mild reaction conditions, and environmental protection, they have gradually become important tools in many fields such as chemical industry, medicine, energy, and materials. For example, microorganisms such as lactic acid bacteria, yeast, and Streptomyces have been successfully used in the synthesis of a variety of biomaterials. However, with the expansion of application fields, the complex relationship between microorganisms and nanomaterials has become increasingly prominent, and researchers are in urgent need of an efficient tool to systematically explore and analyze these relationships.

[0003] In traditional research, the synthesis of nanomaterials by microorganisms mainly relies on the knowledge accumulation of experts and experimental verification. These research methods usually include literature retrieval, experimental verification, data analysis and other steps. Researchers screen out microorganisms that may synthesize nanomaterials by consulting a large amount of scientific research literature, and then verify them through experiments. The main advantage of this method is that its research results are highly reliable, and the experimental data directly supports the correlation between microorganisms and nanomaterials. However, the disadvantages of this method are also very obvious.

[0004] For example, manual retrieval and screening of literature is inefficient. With the growth of the number of scientific research literature, researchers are faced with an increasing amount of information, and it is difficult to comprehensively and systematically review and analyze all research-related literature. This not only increases the workload of researchers, but may also lead to the omission of some potential associations between microorganisms and nanomaterials. Secondly, the experimental verification cycle is long and the cost is high. Each association between microorganisms and nanomaterials needs to be verified experimentally, usually involving complex biological culture, chemical analysis and other processes. These experiments are not only time-consuming, but also require a large amount of experimental materials and equipment. In addition, due to the complexity of experimental conditions, the results between different laboratories may differ, further increasing the uncertainty of the research. Therefore, traditional research methods are difficult to adapt to diverse research needs. Researchers are not only concerned with the direct relationship between microorganisms and nanomaterials, but also want to understand the potential associations between them. These complex associations are difficult to fully reveal through traditional methods.

[0005] In recent years, with the development of big data technology, knowledge graphs have gradually become an important tool for studying complex systems and association mining. Knowledge graphs are a structured way of representing knowledge, which visualizes different entities and their relationships in the form of nodes and edges. Knowledge graphs can not only integrate multi-source heterogeneous data, but also reveal potential associations in complex systems through graph algorithms.

[0006] In the field of bioinformatics, the application of knowledge graphs has gradually deepened. Researchers use knowledge graphs to integrate a large amount of bioinformatics data, such as genomic data, proteomic data, metabolic network data, etc. These data are integrated and analyzed through knowledge graphs, which can provide researchers with a new perspective and reveal potential associations that are difficult to discover with traditional methods. For example, researchers used knowledge graphs to discover the potential association between certain genes and diseases, which promoted the development of personalized medicine.

[0007] In the study of microbial synthesis of nanomaterials, knowledge graphs also have broad application prospects. By constructing a knowledge graph of the relationship between microorganisms and nanomaterials, researchers can systematically integrate multi-source data such as literature and experimental data to structure microorganisms, nanomaterials and their properties and relationships. Knowledge graphs can not only intuitively display the direct relationship between microorganisms and nanomaterials, but also mine the potential relationship between them through graph algorithms.

[0008] Although knowledge graphs have shown great potential in the mining of associations between microorganisms and nanomaterials, their application still faces many challenges. For example, data acquisition and processing are still one of the main difficulties in building knowledge graphs. In the study of microorganisms and nanomaterials, data sources are extensive, including scientific research literature, experimental data, databases, etc. These data are often heterogeneous, and the data format, structure and content are different. How to effectively collect, clean and standardize these data to ensure their accuracy and consistency is the primary task of knowledge graph construction. In addition, how to maintain the simplicity of its structure and the accuracy of information while ensuring the scale of the graph is the key to building a high-quality knowledge graph. With the continuous progress of scientific research, new data and new knowledge continue to emerge, and knowledge graphs need to be continuously updated and maintained to ensure their timeliness and practicality. Association mining of knowledge graphs relies on advanced graph algorithms and machine learning techniques. Traditional graph algorithms, such as path analysis and cluster analysis, can reveal some simple associations in the graph, but for complex biological systems, the capabilities of traditional algorithms are limited. Summary of the invention

[0009] A method for recommending microbial synthesis of nanomaterials based on knowledge graph, comprising:

[0010] Obtain information on microorganisms and nanomaterials;

[0011] A knowledge graph is constructed through the microorganism information and the nanomaterial information; wherein the nodes of the knowledge graph are composed of microorganisms, nanomaterials, synthesis methods and elements;

[0012] Obtaining a structural feature vector and a semantic feature vector based on the knowledge graph;

[0013] The structural feature vector and the semantic feature vector are concatenated to obtain a representation vector of each node; each node is scored based on the representation vector, and the relationship between the microorganism and the nanomaterial is determined according to the scoring result.

[0014] Preferably, in the ontology of the knowledge graph, entity tags include: biology, habitat, mechanism, function, metal resistance, nanomaterial, synthesis method, precursor, pH value, temperature, rotation speed, position, size, shape, application, antimicrobial agent, electronics and energy.

[0015] Preferably, the knowledge graph obtains the structural feature vector and the semantic feature vector through a BERT-RGCN model, wherein the BERT-RGCN model includes a graph structure processing branch and a text data processing branch.

[0016] Preferably, the features of the graph structure are extracted through the graph structure processing branch.

[0017] Preferably, semantic information of the text data is captured by the text data processing branch.

[0018] Preferably, the extracting step comprises:

[0019] Input a multi-relation graph obtained based on the knowledge graph;

[0020] Update the node representation in the multi-relationship graph based on the relationship type;

[0021] The structural feature vector of the node is output based on the updated node representation.

[0022] Preferably, the capturing step comprises:

[0023] Input a text sequence obtained based on the knowledge graph;

[0024] Convert the text sequence into a word vector;

[0025] The word vector outputs the semantic feature vector of the node through the encoder.

[0026] Preferably, the average reciprocal ranking is used as the evaluation indicator of the method.

[0027] The beneficial effects of the present invention are as follows:

[0028] By applying emerging technologies such as deep learning and graph neural networks to the association mining of knowledge graphs, researchers can mine more complex potential associations from large-scale knowledge graphs, such as the implicit associations between microorganisms and potential nanomaterials. Researchers can use knowledge graphs to comprehensively and systematically analyze the complex relationship between microorganisms and nanomaterials, thereby providing theoretical support for the development and optimization of nanomaterials. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of an embodiment of the present invention.

[0030] Figure 2 Flow chart of the model training phase in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] According to the following Figure 1 The embodiments of the present invention are described in detail. The embodiments are implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.

[0032] 1. Get data

[0033] Acquire data. The data sources include at least literature data and NCBI data.

[0034] Literature data at least includes title, microorganism, synthesis method, and nanomaterial, and the above information is manually annotated.

[0035] NCBI data at least includes microbial information and microbial mechanism information.

[0036] The above data provide basic data support for the construction of the knowledge graph. For missing data, it can be obtained by querying LLM.

[0037] 2. Data Preprocessing

[0038] Because the above data come from different sources, the data formats and structures vary greatly.

[0039] Preprocess the acquired data, including data cleaning, data standardization and data integration operations.

[0040] Data cleaning includes: removing duplicate data, processing missing values, and correcting errors, aiming to remove or correct noise, redundancy, and erroneous data in the data.

[0041] In the missing value processing step, different methods are used to handle missing data according to different situations. If the data annotation lacks microorganisms or nanomaterials, the annotated data will be deleted; if the synthesis method is missing, the synthesis method will be directly replaced with a number during processing; if the microbial mechanism is missing, the name node corresponding to the microorganism will be searched for the corresponding microbial mechanism information in the NCBI database. If there is no corresponding microbial mechanism in the NCBI database, LLM will be used to generate the corresponding microbial mechanism.

[0042] Data standardization aims to make data from different sources have a unified format and structure for subsequent map construction. For example, the labeled data "ag", "Ag", "Agg", "silver" all represent the material "silver", and they are unified as "silver".

[0043] Data integration aims to integrate data from different sources to form a unified whole logically. The data integration process includes data mapping, entity alignment, and relationship integration operations.

[0044] 3. Building a Knowledge Graph

[0045] The knowledge graph construction process includes defining nodes, defining edges, selecting entities, selecting relationships, graph construction, and graph storage steps.

[0046] Define nodes, which represent entities in the knowledge graph. In the present invention, nodes are classified as follows:

[0047] Microbial nodes: represent the types of microorganisms involved in the production of synthetic nanomaterials.

[0048] Material node: represents nanomaterials synthesized by microorganisms.

[0049] Synthesis method node: indicates how microorganisms synthesize nanomaterials in a paper.

[0050] Element nodes: Indicates which elements are contained in each nanomaterial.

[0051] Define edges, which represent the relationship between entities. In the present invention, the edges are classified as follows:

[0052] Synthesis relationship: It means that a certain microorganism can synthesize a specific nanomaterial.

[0053] Inclusion relationship: indicates which elements each nanomaterial contains.

[0054] Microorganisms, nanomaterials, synthesis methods, and elements are selected from the above data as entity nodes in the knowledge graph to complete entity selection; the microorganism and material nodes in the same document are connected through synthesis relationships, and the materials are connected to their constituent elements to complete relationship selection.

[0055] After the entities and relationships are selected, they are combined to build a knowledge graph. The constructed graph is a multi-attribute directed graph, with nodes connected by edges. Each node and edge contains multiple attribute information, such as the material node contains shape, size, and location attributes. The knowledge graph can not only display the entities and relationships in the data, but also provide support for subsequent algorithms.

[0056] In order to construct a domain knowledge dataset, the present invention designs an ontology in which entity labels are provided, including data such as microorganisms, habitats, mechanisms, functions, metal resistance, nanomaterials, synthesis methods, precursors, pH values, temperatures, rotation speeds, positions, sizes, shapes, applications, antimicrobial agents, electronics, and energy.

[0057] Graph storage. The constructed knowledge graph needs to be stored for subsequent query and analysis. In the present invention, Neo4j graph database is used for storage.

[0058] 4. Model Training

[0059] In order to process graph structure information and text information, the present invention designs a BERT-RGCN model, such as Figure 2 As shown in the figure, it includes two branches: graph structure processing branch and text processing branch. These two branches are responsible for processing graph data and text data respectively, and fusing their features in subsequent layers.

[0060] 1. Graph structure processing branch

[0061] RGCN is used as the basic model to extract the features of the graph structure. Pooling layer: Pooling the feature map output by the convolutional layer to reduce the feature dimension and improve the robustness of the model. Fully connected layer: Connect the output of the pooling layer to one or more fully connected layers to learn higher-level image feature representations.

[0062] The RGCN model is a neural network model specifically designed to process multi-relational graph data. It can capture the complex relationships between nodes and their neighboring nodes in the graph through convolution operations. The RGCN model takes nodes and their connection relationships as input, and gradually updates the node representation vector through multiple layers of convolution operations. Each layer of convolution operations can further extract and integrate the local neighborhood information of the node, so that the final output node representation can fully reflect its position and role in the entire graph structure. Through training, the RGCN model can generate structural feature vectors for each node, which represent the local and global information of the node in the graph structure. Since the RGCN model captures the dependencies between nodes through relational convolution, it is particularly suitable for processing knowledge graph data with complex relationships.

[0063] The node representation vector generated by the RGCN model is further transformed by nonlinear transformation (ReLU activation function) and linear transformation. Through these steps, the structural features of the node are mapped to a new feature space for integration with subsequent semantic features. This stage of processing helps the model to more effectively fuse feature information from different sources in subsequent steps.

[0064] RGCN is an extension of Graph Neural Networks (GNN) for processing heterogeneous graphs, i.e. graphs with multiple different types of nodes and edges. A multi-relation graph is a complex graph structure in which nodes of different types are connected by multiple relationship types. In standard graph convolutional networks (GCN), the type information of edges is usually ignored, while RGCN retains the type information of edges by introducing relationship matrices and corresponding transformations, thereby better capturing the rich semantic relationships in the graph. The basic structure of RGCN is similar to that of standard GCN, but the core difference is the introduction of transformation matrices for different relationship types. The input of the RGCN model is a multi-relation graph, represented as:

[0065] G=(V,E,R)

[0066] Where V is the node set, E is the edge set, and R is the relationship set. Each layer of RGCN updates the representation of the node, and this update process depends on the node's neighbor nodes and their connection relationships. For each relationship type, RGCN defines an independent linear transformation for it, and then combines information from different relationship types to update the node representation. The key to RGCN is how to effectively update the node representation in a multi-relationship graph. Assume that the node set in the graph is v, the edge set is ε, and the relationship set is R. For the representation of node v At layer l, RGCN is updated by aggregating information from neighboring nodes. The specific update formula is:

[0067]

[0068] in represents the set of neighbor nodes connected to node v through relationship r; is the transformation matrix of relation r at level l; The feature vector of node u in layer l; is the transformation matrix of the node’s own information; c v,r is a normalization constant used to adjust the contribution of different relationship types to the node representation update; σ(·) is a nonlinear activation function, such as ReLU.

[0069] In RGCN, different types of relations may contribute differently to the update of node representation. By applying different weights to each relation type, RGCN can flexibly aggregate information from different relations. These weights are transformed through the transformation matrix W. r ,accomplish:

[0070]

[0071] This formula shows how to perform a weighted sum of information from different relationship types and perform a nonlinear transformation on node representations via an activation function.

[0072] Since the number of parameters in the RGCN model increases with the increase of relation types, it may lead to overfitting, especially when there is less training data. To solve this problem, RGCN introduces edge-type-based regularization by replacing the transformation matrix W r Decompose into a product of low-rank matrices to reduce model parameters:

[0073] W r =W base +A r B r

[0074] Where W base is the shared base transformation matrix, A r and B r is a matrix specific to relation r.

[0075] 2. Text data processing branch

[0076] The BERT model receives the text description of the node, processes the text through its multi-layer Transformer encoder, and generates a semantic feature vector corresponding to the node to supplement the structural features generated by the RGCN model.

[0077] BERT (Bidirectional Encoder Representations from Transformers) is a pre-trained language model that significantly improves the performance of natural language processing (NLP) tasks through bidirectional encoders and deep learning techniques. The core of BERT is based on the Transformer architecture, which is pre-trained with a large amount of unsupervised text data to generate context-sensitive word vector representations, thereby performing well in downstream tasks.

[0078] BERT's architecture is based on the encoder part of the Transformer, a deep neural network structure based entirely on the attention mechanism. The input to the BERT model is a text sequence that has been segmented, which is converted into word vectors and added to the positional encoding to preserve the position information in the sequence. The basic unit of the Transformer is a multi-head self-attention mechanism and a feedforward neural network. BERT builds a deep model by stacking multiple such basic units (called "layers"). In the standard version of BERT, a 12-layer encoder is used, each layer contains 768 hidden units, and the total number of parameters exceeds 100 million.

[0079] The core mathematical formulas in BERT mainly include multi-head self-attention mechanism, feedforward neural network and position encoding. The self-attention mechanism is used to calculate the dependency between each word and other words in the input sequence. Given the word vector representation of the input sequence:

[0080] X=[x1,x2,x3,...,x n ]

[0081] The calculation of self-attention is as follows:

[0082] Query, Key, Value means:

[0083] Q=XW Q

[0084] K=XW K

[0085] V=XW V

[0086] Among them, W Q , W K , W V is a trainable weight matrix. The calculation formula of the attention score is as follows:

[0087]

[0088] Where dk is the dimension of K. Multi-head attention calculates attention through multiple different heads, concatenates the results and maps them back to the original dimension:

[0089] MultiHead(Q,K,V)=Concat(head1,…,head h )W O

[0090] After the self-attention mechanism, the input is processed by a feed-forward neural network. The form of the feed-forward network at each layer is:

[0091] FFN(x)=ReLU(xW1+b1)W2+b2

[0092] Among them, W1, W2, b1, b2 are trainable parameters.

[0093] 3. Vector stitching

[0094] The structural feature vector of the RGCN model and the semantic feature vector from the BERT model are concatenated. In this way, the model can retain both the structural information and semantic information of the node, thereby generating a comprehensive node representation vector. It can not only reflect the structural position of the node in the knowledge graph, but also reflect the content and semantics of the node. For example, the vector of the microorganism node is m1, and the vector of the nanomaterial node is m2. The scoring formula for the microorganism and nanomaterial nodes is as follows:

[0095]

[0096] For a given nanomaterial, each microorganism is scored using the above scoring formula, and then the score is sorted. During the training process, the present invention uses MRR (mean reciprocal rank) as an evaluation indicator. MRR is used to measure the effectiveness of the information retrieval system or model, especially focusing on the quality of the first relevant result returned. The value of MRR is between 0 and 1. The larger the value, the better the model ranks the relevant results in the front. Specifically, MRR measures the overall performance of the system for all queries by calculating the reciprocal of the ranking of the first relevant result of each query, and then averaging the results of all queries. The calculation formula is as follows:

[0097]

[0098] Where |Q| represents the total number of query sets. i Represents the ranking position of the first relevant result for the i-th query.

[0099] V. Examples of achievements

[0100] Using some of the invention achievements that can be disclosed as examples, for a given nanomaterial, the present invention predicts candidate microbial entities and obtains the conclusion that "the microorganism Shewanella oneidensis MR-1 can synthesize the nanomaterial AuPdPt", which has been verified by experts in the field.

[0101] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. A person skilled in the art may modify or make equivalent substitutions for the technical solutions of the present invention without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. A method for recommending microbial synthesis of nanomaterials based on knowledge graph, comprising: Obtain information on microorganisms and nanomaterials; A knowledge graph is constructed through the microorganism information and the nanomaterial information; wherein the nodes of the knowledge graph are composed of microorganisms, nanomaterials, synthesis methods and elements; Obtaining a structural feature vector and a semantic feature vector based on the knowledge graph; Concatenating the structural feature vector and the semantic feature vector to obtain a representation vector of each node; Scoring each of the nodes based on the representation vector, and determining the relationship between the microorganism and the nanomaterial according to the scoring result.

2. The method according to claim 1, characterized in that In the ontology of the knowledge graph, entity tags include: organism, habitat, mechanism, function, metal resistance, nanomaterial, synthesis method, precursor, pH value, temperature, rotation speed, position, size, shape, application, antimicrobial agent, electronics and energy.

3. The method according to claim 1, characterized in that The knowledge graph obtains the structural feature vector and the semantic feature vector through the BERT-RGCN model, wherein the BERT-RGCN model includes a graph structure processing branch and a text data processing branch.

4. The method according to claim 3, characterized in that The features of the graph structure are extracted through the graph structure processing branch.

5. The method according to claim 3, characterized in that: Semantic information of text data is captured through the text data processing branch.

6. The method according to claim 4, characterized in that The extraction steps include: Input a multi-relation graph obtained based on the knowledge graph; Update the node representation in the multi-relationship graph based on the relationship type; The structural feature vector of the node is output based on the updated node representation.

7. The method according to claim 5, characterized in that The capturing steps include: Input a text sequence obtained based on the knowledge graph; Convert the text sequence into a word vector; The word vector outputs the semantic feature vector of the node through the encoder.

8. The method according to claim 1, characterized in that The average reciprocal rank is used as the evaluation indicator of the method.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, the computer program is configured to be executed by the processor, and the computer program comprises instructions for executing the method according to any one of claims 1 to 8.

10. A storage medium storing a computer program, wherein when the computer program is executed by a computer, the method according to any one of claims 1 to 8 is implemented.

Images