Establishment method of acute toxicity prediction model of multi-species aquatic organisms and toxicity prediction method

Abstract

The invention provides an establishment method of a multi-species aquatic organism acute toxicity prediction model and a toxicity prediction method, and belongs to the technical field of environmental science and ecotoxicology. The establishment method of the acute toxicity prediction model of the multi-species aquatic organisms comprises the steps that actual acute toxicity data of multiple compounds on the multi-species aquatic organisms, molecular maps of the multiple compounds and species information of the multi-species aquatic organisms are obtained, and the species information comprises Cytb common subsequences and / or species chemical information of the multi-species aquatic organisms; inputting the molecular maps of the various compounds and the species information of the multi-species aquatic organisms into an initial prediction model to obtain initial acute toxicity prediction data of the various compounds on the multi-species aquatic organisms; and optimizing network parameters of the initial prediction model according to the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

Description

Technical Field

[0001] The present disclosure relates to the technical field of environmental science and ecotoxicology, and more specifically, to a method for establishing a multi-species aquatic organism acute toxicity prediction model and a toxicity prediction method. Background Art

[0002] In the field of environmental science and ecotoxicology, the health status of aquatic ecosystems is widely considered to be an important indicator of environmental quality and pollution levels. In particular, aquatic organisms, due to their high sensitivity to environmental changes and unique ecological adaptability, have become key biological indicators for monitoring environmental disturbances, especially ecological disturbances caused by industrial chemicals and agricultural chemicals. In the risk assessment of ecotoxicology, acute toxicity testing is a core indicator for evaluating the effects of chemicals on organisms, usually involving the assessment of the impact on the mortality, reproductive capacity, mobility or growth status of specific species.

[0003] With the rapid growth of chemical products, more than 350,000 chemicals have been registered and used worldwide, which poses a potential threat to human health and environmental safety. There is an urgent need for a comprehensive and systematic assessment of these potential environmental pollutants. However, traditional in vivo toxicology experiments are not only costly and time-consuming, but also rely on a large number of experimental animals. Summary of the invention

[0004] In view of this, the present disclosure provides a method for establishing a multi-species aquatic organism acute toxicity prediction model and a toxicity prediction method.

[0005] One aspect of the present disclosure provides a method for establishing a multi-species aquatic organism acute toxicity prediction model, comprising: obtaining actual acute toxicity data of multiple compounds to multiple species aquatic organisms, molecular graphs of the multiple compounds and species information of the multiple species aquatic organisms, wherein the species information includes Cytb common subsequences and / or species evolution information of the multiple species aquatic organisms; inputting the molecular graphs of the multiple compounds and the species information of the multiple species aquatic organisms into an initial prediction model to obtain initial acute toxicity prediction data of the multiple compounds to the multiple species aquatic organisms; optimizing the network parameters of the initial prediction model based on the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

[0006] Another aspect of the present disclosure provides an acute toxicity prediction method, comprising: obtaining a molecular graph of a compound to be predicted and species information of an organism to be predicted, wherein the species information includes a Cytb common subsequence and / or species evolution information of the organism to be predicted; inputting the molecular graph of the compound to be predicted and the species information of the organism to be predicted into a multi-species aquatic organism acute toxicity prediction model to obtain acute toxicity prediction data of the compound to be predicted to the organism to be predicted; wherein the multi-species aquatic organism acute toxicity prediction model is obtained by the above-mentioned establishment method.

[0007] Another aspect of the present disclosure provides a device for establishing a multi-species aquatic organism acute toxicity prediction model, including: a first acquisition module, used to obtain actual acute toxicity data of multiple compounds on multiple species aquatic organisms, molecular graphs of the above-mentioned multiple compounds and species information of the above-mentioned multiple species aquatic organisms, wherein the above-mentioned species information includes Cytb common subsequences and / or species evolution information of the above-mentioned multiple species aquatic organisms; a first acquisition module, used to input the molecular graphs of the above-mentioned multiple compounds and the species information of the above-mentioned multiple species aquatic organisms into the initial prediction model to obtain the initial acute toxicity prediction data of the above-mentioned multiple compounds on the above-mentioned multiple species aquatic organisms; an optimization module, used to optimize the network parameters of the above-mentioned initial prediction model according to the above-mentioned initial acute toxicity prediction data and the above-mentioned actual acute toxicity data, to obtain a trained multi-species aquatic organism acute toxicity prediction model.

[0008] Another aspect of the present disclosure provides an acute toxicity prediction method, including: a second acquisition module, used to obtain the molecular graph of the compound to be predicted and the species information of the organism to be predicted, wherein the above-mentioned species information includes the Cytb common subsequence and / or species evolution information of the above-mentioned organism to be predicted; a second acquisition module, used to input the molecular graph of the above-mentioned compound to be predicted and the species information of the above-mentioned organism to be predicted into a multi-species aquatic organism acute toxicity prediction model to obtain the acute toxicity prediction data of the above-mentioned compound to be predicted to the above-mentioned organism to be predicted; wherein the above-mentioned multi-species aquatic organism acute toxicity prediction model is obtained by the method established by the above-mentioned method.

[0009] Another aspect of the present disclosure provides an electronic device, comprising:

[0010] one or more processors;

[0011] a memory for storing one or more programs,

[0012] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described above.

[0013] Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above method when executed.

[0014] Another aspect of the present disclosure provides a computer program product, which includes computer executable instructions, and when the instructions are executed, are used to implement the method as described above.

[0015] According to the embodiments of the present disclosure, the molecular graphs of multiple compounds and the species information of multiple species of aquatic organisms are used as inputs of the initial prediction model, and the actual acute toxicity data of known compounds on multiple species of aquatic organisms are used as labels to train the initial prediction model, thereby better learning the molecular characteristics of the acute toxic effects of different compounds on different species of aquatic organisms, thereby establishing a multi-species aquatic organism acute toxicity prediction model that can target multiple compounds and multiple species of aquatic organisms, and has more extensive and convenient applications and prediction performance.

[0016] According to the embodiments of the present disclosure, species information including Cytb common subsequences and / or species evolution information of multiple species of aquatic organisms can be used to better extract information between species. Among them, Cytb common subsequences have relatively stable characteristics, and species evolution information can better perform cluster analysis on species.

[0017] The acute toxicity prediction method provided in the embodiments of the present disclosure is suitable for predicting the acute toxicity of large-scale compounds and multi-species aquatic organisms. The prediction method is simple and efficient, improves the prediction performance, expands the prediction range, and has broad application prospects in the fields of chemical environmental health risk assessment, environmental assessment, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0019] Figure 1 The exemplary system architecture of the method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method and device thereof according to the embodiment of the present disclosure is schematically shown;

[0020] Figure 2 A flowchart schematically shows a method for establishing a multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure;

[0021] Figure 3 The schematic diagram of the structure of the multi-species aquatic organism acute toxicity prediction model according to the embodiment of the present disclosure is shown;

[0022] Figure 4The method for obtaining initial acute toxicity prediction data according to an embodiment of the present disclosure is schematically shown;

[0023] Figure 5 A flowchart schematically shows a method for establishing a multi-species aquatic organism acute toxicity prediction model according to another embodiment of the present disclosure;

[0024] Figure 6 A flowchart of a method for predicting acute toxicity of multiple species of aquatic organisms according to an embodiment of the present disclosure is schematically shown;

[0025] Figure 7 A block diagram schematically shows a device for establishing a multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure;

[0026] Figure 8 A block diagram schematically shows a device for establishing a multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure; and

[0027] Fig. 9 A block diagram of an electronic device suitable for implementing a method for establishing a multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0028] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0029] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0030] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0031] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0032] In the embodiments of the present disclosure, the collection, updating, analysis, processing, use, transmission, provision, disclosure, storage, etc. of the data involved (for example, including but not limited to user personal information) are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain user personal information security and network security. In the embodiments of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting user personal information.

[0033] Traditional in vivo toxicology experiments are not only costly and time-consuming, but also rely on a large number of experimental animals. To address these issues, in vitro computational methods can be considered to replace or reduce traditional in vivo experiments. These computational methods can more quickly and accurately predict the potential toxicity of chemicals to aquatic organisms, providing more effective risk assessment and management strategies for environmental protection and human health.

[0034] In related technologies, the prediction model established using traditional machine learning algorithms requires humans to use quantum chemical calculation methods to calculate the quantitative parameters of molecules as molecular descriptors. The preliminary preparation of data will consume a lot of time and computing resources. In addition, the prediction model established by traditional machine learning algorithms is only established for a single species, with a narrow application range and limited prediction capabilities, and the prediction performance needs to be improved.

[0035] Quantitative Structure-Activity Relationship (QSAR) links the biological effects of chemicals to their molecular structure and properties, becoming the key to guiding the prediction of molecular properties. QSAR is based on a key assumption that molecules with similar structures usually exhibit similar functional properties. By analyzing the relationship between molecular structure and properties, QSAR models can quickly and efficiently predict the properties of molecules without laboratory experiments.

[0036] A major challenge in building an efficient prediction model is the scarcity and imbalance of aquatic toxicity data. Data from toxicity databases show that high-quality aquatic toxicity data are scarce and unevenly distributed. Aquatic toxicity data are mainly concentrated on fish, while toxicity data for other species such as algae and crustaceans are very limited. This uneven data distribution hinders the establishment of a comprehensive aquatic toxicity prediction model. In addition, due to the differences in toxicity test conditions for different species, data utilization is difficult. In the face of these challenges, there is an urgent need to develop more accurate and comprehensive aquatic toxicity prediction models to more accurately assess the potential toxicity of chemicals to various aquatic organisms. These models need to cover not only common aquatic species, such as Daphnia, algae, and fish as ecological indicator species, but also a variety of other aquatic organisms. In this way, stronger scientific support can be provided for environmental protection, while also providing important data support for the safe use of chemicals.

[0037] In view of this, an embodiment of the present disclosure provides a method for establishing a multi-species aquatic organism acute toxicity prediction model, including: obtaining actual acute toxicity data of multiple compounds on multiple species aquatic organisms, molecular graphs of multiple compounds and species information of multiple species aquatic organisms, wherein the species information includes Cytb common subsequences and / or species evolution information of multiple species aquatic organisms; inputting the molecular graphs of multiple compounds and the species information of multiple species aquatic organisms into an initial prediction model to obtain initial acute toxicity prediction data of multiple compounds on multiple species aquatic organisms; optimizing the network parameters of the initial prediction model based on the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

[0038] Figure 1 The exemplary system architecture 100 of the method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method and device thereof according to the embodiment of the present disclosure is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0039] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0040] The user may use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only for example).

[0041] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0042] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0043] It should be noted that the method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the device for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction device provided in the embodiment of the present disclosure can generally be set in the server 105. The method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the device for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Alternatively, the method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method provided in the embodiment of the present disclosure may also be performed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may also be performed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Accordingly, the device for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction device provided in the embodiment of the present disclosure may also be provided in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may be provided in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0044] For example, the actual acute toxicity data of multiple compounds to multiple species of aquatic organisms, the molecular graphs of multiple compounds, and the species information of multiple species of aquatic organisms may be originally stored in any one of the first terminal device 101, the second terminal device 102, or the third terminal device 103 (for example, the first terminal device 101, but not limited thereto), or stored in an external storage device and may be imported into the first terminal device 101. Then, the first terminal device 101 may locally execute the method for establishing a multi-species aquatic organism acute toxicity prediction model and the toxicity prediction method provided in the embodiment of the present disclosure, or send the actual acute toxicity data of multiple compounds to multiple species of aquatic organisms, the molecular graphs of multiple compounds, and the species information of multiple species of aquatic organisms to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the actual acute toxicity data of multiple compounds to multiple species of aquatic organisms, the molecular graphs of multiple compounds, and the species information of multiple species of aquatic organisms may execute the method for establishing a multi-species aquatic organism acute toxicity prediction model provided in the embodiment of the present disclosure.

[0045] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0046] Figure 2 The flowchart of the method for establishing a multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure is schematically shown.

[0047] like Figure 2 As shown, the method includes operations S100 to S300.

[0048] In operation S100, actual acute toxicity data of multiple compounds on multiple species of aquatic organisms, molecular graphs of multiple compounds, and species information of multiple species of aquatic organisms are obtained, wherein the species information includes Cytb common subsequences and / or species evolution information of multiple species of aquatic organisms.

[0049] In operation S200, the molecular graphs of the plurality of compounds and the species information of the plurality of aquatic organisms are input into an initial prediction model to obtain initial acute toxicity prediction data of the plurality of compounds to the plurality of aquatic organisms.

[0050] In operation S300, the network parameters of the initial prediction model are optimized according to the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

[0051] According to an embodiment of the present disclosure, the plurality of compounds may be any compounds. For example, they may be compounds that are toxic to aquatic organisms. Specifically, the plurality of compounds may be halogenated phenol compounds, nitrobenzene compounds, alkylbenzene compounds, and the like.

[0052] According to an embodiment of the present disclosure, the multi-species aquatic organisms may include, for example, aquatic organisms of various species such as algae, fish, and crustaceans.

[0053] According to an embodiment of the present disclosure, the molecular graph of a compound may be an undirected graph including node features and edge features, wherein the node features are used to characterize atomic features, and the edge features are used to characterize bond features.

[0054] According to an embodiment of the present disclosure, species information may include Cytb common subsequences of multiple species of aquatic organisms; species information may include species evolution information of multiple species of aquatic organisms; species information may also include Cytb common subsequences and species evolution information of multiple species of aquatic organisms.

[0055] According to an embodiment of the present disclosure, Cytb (Cytochrome b gene) is a cytochrome b gene. The Cytb sequence information of different aquatic organisms can be obtained, and the MEGA11 software is used for matching, alignment, and trimming to obtain the Cytb common subsequence as the species information of multiple species of aquatic organisms.

[0056] According to an embodiment of the present disclosure, the species evolution information includes species classification information of multiple species of aquatic organisms. The species classification information of organisms may be information such as the kingdom, phylum, class, order, family, genus, and species to which the organisms belong.

[0057] According to the embodiments of the present disclosure, the acute toxicity data of the compound to organisms can be measured using LC 50 Measure, LC 50 It refers to the median lethal concentration, which is the concentration of a compound that kills half of the species. The higher the concentration, the lower the toxicity.

[0058] According to the embodiments of the present disclosure, the molecular graphs of multiple compounds and the species information of multiple species of aquatic organisms are used as inputs of the initial prediction model, and the actual acute toxicity data of known compounds on multiple species of aquatic organisms are used as labels to train the initial prediction model, thereby better learning the molecular characteristics of the acute toxic effects of different compounds on different species of aquatic organisms, thereby establishing a multi-species aquatic organism acute toxicity prediction model that can target multiple compounds and multiple species of aquatic organisms, and has more extensive and convenient applications and prediction performance.

[0059] According to the embodiments of the present disclosure, by including the common subsequences of Cytb and / or species evolution information of multiple species of aquatic organisms, information between species can be better extracted. Among them, the common subsequences of Cytb have relatively stable characteristics, and the species evolution information can better perform cluster analysis on species.

[0060] According to the embodiments of the present disclosure, the molecular graph of the compound is used as the input of the initial prediction model, and there is no need to use the molecular weight quantitative parameters calculated using quantum chemical calculation methods in traditional machine learning as molecular descriptors, thereby saving time and computing resources for molecular description calculation and selection, and reducing the requirements for computational chemistry foundations when applying.

[0061] Reference below Figure 3~Figure 5 , combined with specific embodiments Figure 2 The method shown is further explained.

[0062] Figure 3 The structural diagram of the multi-species aquatic organism acute toxicity prediction model according to an embodiment of the present disclosure is schematically shown.

[0063] According to the embodiments of the present disclosure, Figure 3As shown, the initial prediction model includes a feature extraction network 310, an interaction layer 320 and a prediction layer 330 connected in sequence, and the feature extraction network 310 includes a species feature extraction network 311 and a molecular feature extraction network 312.

[0064] According to an embodiment of the present disclosure, deep learning is a type of machine learning algorithm that can use hierarchical reorganization of features to extract relevant information and learn the pattern properties represented by data.

[0065] Figure 4 The method for obtaining initial acute toxicity prediction data according to an embodiment of the present disclosure is schematically illustrated.

[0066] like Figure 4 As shown, the molecular graphs of multiple compounds and the species information of multiple species of aquatic organisms are input into the initial prediction model to obtain the initial acute toxicity prediction data of the multiple compounds to multiple species of aquatic organisms, including operations S210~S240.

[0067] In operation S210, the species information 301 of the multi-species aquatic organisms is input into the species feature extraction network 311, and the species feature tensor of the species information of the multi-species aquatic organisms is extracted.

[0068] In operation S220 , the molecular graphs 302 of the plurality of compounds are input to the molecular feature extraction network 312 to extract molecular feature tensors of the molecular graphs of the plurality of compounds.

[0069] In operation S230 , the species feature tensor and the molecule feature tensor are input to the interaction layer 320 to obtain a combined feature tensor of the species feature tensor and the molecule feature tensor.

[0070] In operation S240 , the combined feature tensor is input to the prediction layer 330 to obtain initial acute toxicity prediction data.

[0071] According to the embodiments of the present disclosure, the species feature tensor of the species information of multi-species aquatic organisms can be extracted using a feature extraction network; the molecular feature tensor can be extracted using the molecular graphs of multiple compounds; the species feature tensor and the molecular feature tensor are combined using an interaction layer to obtain a combined feature tensor; the combined feature tensor is predicted using a prediction layer to obtain initial acute toxicity prediction data. Thus, the species information of multi-species aquatic organisms and the molecular graphs of multiple compounds are sequentially subjected to feature extraction and feature combination, and then an initial acute toxicity prediction is performed, so that the initial prediction model can better learn the acute toxic effects of different compounds on different aquatic species, so that the trained multi-species aquatic organism acute toxicity prediction model has a larger prediction range and higher prediction efficiency.

[0072] According to an embodiment of the present disclosure, the species information may include Cytb common subsequences of multiple species of aquatic organisms. Operation S210 may include: extracting features from the Cytb common subsequences to obtain a Cytb feature tensor, which is a species feature tensor.

[0073] According to an embodiment of the present disclosure, the species information may also include species evolution information of multiple species of aquatic organisms. Operation S210 may include: extracting features from the species evolution information to obtain a species evolution tensor, which is a species feature tensor.

[0074] According to an embodiment of the present disclosure, species information may also include Cytb common subsequences and species evolution information of multiple species of aquatic organisms.

[0075] According to an embodiment of the present disclosure, operation S210, that is, inputting the species information 301 of multiple species of aquatic organisms into the species feature extraction network 311 to extract the species feature tensor of the species information of multiple species of aquatic organisms, may include operations S211 to S213.

[0076] In operation S211, feature extraction is performed on the Cytb common subsequence to obtain a Cytb feature tensor.

[0077] In operation S212, feature extraction is performed on the species evolution information to obtain a species evolution tensor.

[0078] In operation S213, the Cytb feature tensor and the species evolution tensor are concatenated to obtain a species feature tensor.

[0079] According to the embodiments of the present disclosure, by including the species information of the Cytb common subsequences and species evolution information of multiple species of aquatic organisms, the information between species can be better extracted. Among them, the Cytb common subsequences are often used to construct biological evolution trees and have relatively stable characteristics; the species evolution information can better cluster the species, but ignores the microscopic differences in species kinship. Using the species information of the Cytb common subsequences and species evolution information of multiple species of aquatic organisms at the same time, the two are used as a complement to fully extract the species characteristic tensor of the species.

[0080] According to an embodiment of the present disclosure, operation S211, i.e., extracting features from the Cytb common subsequence to obtain a Cytb feature tensor, comprises: one-hot encoding the Cytb common subsequences of multiple species of aquatic organisms in base order to obtain a Cytb feature tensor; and extracting features from the Cytb feature tensor using a convolution kernel to obtain a Cytb feature tensor.

[0081] According to an embodiment of the present disclosure, the ATCG characters in the Cytb sequence can be encoded as [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,01]. Then, the evolutionary classification information of different aquatic organisms is obtained, all of which are summarized and converted into a 01 string, and a value of 1 or 0 is assigned according to whether the species meets the corresponding evolutionary classification group.

[0082] According to an embodiment of the present disclosure, the convolution kernel may be a 4×n×1 convolution kernel, specifically, for example, a 4×3×1 convolution kernel.

[0083] According to an embodiment of the present disclosure, operation S212, i.e., extracting features from species evolution information to obtain a species evolution tensor, includes: performing one-hot encoding on species classification information to obtain a species evolution tensor; and using a fully connected layer to extract features from the species evolution tensor to obtain the species evolution tensor.

[0084] According to the embodiments of the present disclosure, evolutionary populations of multiple species of aquatic organisms can be collected and represented using a character string equal to the length of their number, with 1 assigned to the corresponding positions of the seven evolutionary populations of kingdom, phylum, class, order, family, genus, and species to which the organisms belong, and 0 assigned to the remaining positions.

[0085] According to an embodiment of the present disclosure, the molecular graphs of various compounds include node features for characterizing atomic features and edge features for characterizing bond features.

[0086] According to an embodiment of the present disclosure, the molecular graphs of multiple compounds are established by the following method: according to the SMILES encoding of the multiple compounds, node features for characterizing atomic feature information and edge features for characterizing bond feature information are added to the undirected graph to obtain the molecular graphs of the multiple compounds.

[0087] According to an embodiment of the present disclosure, the atomic feature information includes one or more of the atomic symbol type, atomic implicit valence, atomic free electron number, atomic connection number, atomic formal charge, atomic hybridization mode, atomic connection hydrogen atom number, atomic chirality, and FCFP fingerprint features.

[0088] According to an embodiment of the present disclosure, the bond characteristic information includes one or more of the type of bond, whether the bond is conjugated, whether the bond is on a ring, and the stereo configuration information of the bond.

[0089] According to the embodiments of the present disclosure, SMILES coding is a coding method for describing molecular structures. Molecules are represented by simple symbols of chemical structures such as element symbols, plus and minus signs, brackets, etc., which can quickly and directly represent the topological structure and connection relationship of molecules and can be easily recognized and processed by computer systems. However, features learned directly with SMILES coding will lose a lot of information. The present disclosure converts SMILES coding into molecular graphs, which can better characterize molecular structures, thereby more efficiently extracting key information and reducing unnecessary repetitive steps in calculations. Through the conversion between SMILES coding and molecular graphs, it is more conducive to using the initial prediction model to learn the molecular characteristics of the acute toxicity of compounds to multiple species of aquatic organisms, thereby saving calculation time.

[0090] Specifically, the molecular graph converted from the SMILES encoding of the compound is used as the input of the initial prediction model, and the actual acute toxicity data of known compounds and multi-species aquatic organisms are used as labels to train the initial prediction model and extract the molecular characteristics of the acute toxic effects of the compounds. The multi-species aquatic organism acute toxicity prediction model established in this way is suitable for the prediction of acute toxicity of large-scale compounds and multi-species aquatic organisms, reducing the requirements for computational chemistry foundation when applying the prediction model.

[0091] According to an embodiment of the present disclosure, a molecular graph is a set of nodes and edges, where the nodes correspond to atoms in the molecules and the edges correspond to bonds in the molecules. Each node has a node feature for characterizing atomic features and each edge has an edge feature for characterizing bond features.

[0092] According to an embodiment of the present disclosure, an undirected graph G(V, E) may be created using PyG (graph neural network library, PyTorchGeometric), where V represents an atom set and E represents a bond set.

[0093] Specifically, the molecular graphs of various compounds are established by the following method: first, the RDKit toolkit can be used to obtain the number N of non-hydrogen atoms in the SMILES code of volatile organic compounds, and N nodes are added to the undirected graph G. For the non-hydrogen atoms corresponding to each node, the RDKit toolkit is used to obtain the atomic feature information, and a matrix of size 1×L is encoded using a one-hot encoding method; secondly, the node feature matrices corresponding to all nodes are connected to generate an N×L matrix to store the node features of the molecular graph G, where L is set as needed; thirdly, all combinations of non-hydrogen atoms in the SMILES code of volatile organic compounds are traversed, and edges are added to the non-hydrogen atoms corresponding to all nodes in the molecular graph G. For the chemical bonds represented by each edge, the RDKit toolkit is used to obtain the bond feature information, and a matrix of size 1×K is encoded using a one-hot encoding method; finally, the edge feature matrices corresponding to all edges are connected to generate a 2M×K matrix to store the edge features of the molecular graph G, where M is the number of chemical bonds and K can be set as needed.

[0094] In order to better illustrate the process of converting SMILES code into molecular graph, the volatile organic compound isoprene is taken as an example. The SMILES code of isoprene is C=CC(=C)C.

[0095] For example, the atomic feature information obtained for isoprene includes: atomic symbol type (10 bits), atomic implicit valence (2 bits), atomic free electron number (2 bits), atomic connection number (7 bits), atomic formal charge (3 bits), atomic hybridization type (4 bits), functional class fingerprint (FCFP) features (6 bits), number of hydrogen atoms connected to the atom (5 bits), atomic chirality (2 bits), and whether the atom is chiral (1 bit). Then the first carbon atom feature in the SMILES code of isoprene is [True, False, False, False, False, False, False, False, False, False, False, True, True, False, False, False, False, False, True, False, False, True, False, False, 0, 0, 0, 0, 0, 0, False, True, False, False, False, True, 0]. Use one-hot encoding to generate a matrix of size 1×L=1×42 [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0]. Similarly, connect the node feature matrices corresponding to all nodes to generate a matrix of size N×L=5×42, which is stored as the node features of the molecular graph G.

[0096] For example, the bond feature information obtained for isoprene includes: bond type (4 bits), whether the bond is conjugated (1 bit), whether the bond is on the ring (1 bit), and bond stereo configuration information (4 bits). The feature of the first edge (i.e., C=C bond) added to the first carbon atom in the SMILES encoding of isoprene is [False, True, False, False, True, False, True,False, False, False]. One-hot encoding is used to generate a matrix of size 1×K=1×10 [0, 1, 0, 0,1, 0, 1, 0, 0, 0]. The edge feature matrices corresponding to all edges are connected to generate a 2M×K=8×10 matrix to store the edge features of the molecular graph G.

[0097] According to the embodiments of the present disclosure, the molecular graph converted based on SMILES encoding contains atomic feature information and bond feature information, which can better characterize the molecular structure of the compound, thereby facilitating the subsequent graph neural network to learn the molecular features of the acute toxic effects of the compound on aquatic organisms.

[0098] According to an embodiment of the present disclosure, in operation S220, the molecular feature extraction network is a graph isomorphism network. A graph isomorphism network is a general framework for supervised learning of graph structured data. By extracting the molecular features of a compound through a graph isomorphism network, the correlation between each atom and its neighboring atoms and / or bonds can be better learned, thereby facilitating the improvement of the accuracy of the acute toxicity of the compound to aquatic organisms.

[0099] Graph isomorphism network is a type of graph neural network. Graph neural network is a method based on deep learning. It learns information represented by graphs. It is particularly suitable for research fields such as protein-protein interaction networks and molecular properties that use graphs and networks to express information. Deep learning methods require molecular structures to be encoded and represented mathematically, including graph-based representations and molecular descriptor-based representations, which correspond to graph neural network models and traditional models, respectively. Benchmark tests on multiple data sets show that graph-based models outperform other methods on most data sets and have obvious advantages. There is great potential in using deep learning methods based on graph neural networks to predict the physical and chemical properties of molecules and the toxicity of molecular compounds.

[0100] According to an embodiment of the present disclosure, operation S220 specifically includes: for each node in the molecular graph of the compound, updating the hidden state of each node feature to obtain the updated atomic feature corresponding to each node; and merging the updated atomic features corresponding to all nodes in the molecular graphs of multiple compounds to obtain a molecular feature tensor.

[0101] Specifically, each node The initial feature is set as the node feature , for each time step (or layer) , update the features of each node by aggregating neighbor nodes :

[0102] (1)

[0103] Among them, MLP is a fully connected layer, which is used to perform nonlinear transformation on the aggregated features of nodes. is a learnable parameter used to adjust the effect of self-loops and ensure that each node feature is not completely smoothed during the aggregation process. After T time steps (or layers), the final feature of the node is Combines its own initial features and all neighbor information. For each node The initial features Added to the final hidden state, To obtain the final characteristics of the node , the characteristics of each node The feature tensor A of the entire molecule is obtained by summarizing, which represents the molecular features of the entire compound and can be used for prediction tasks.

[0104] According to an embodiment of the present disclosure, operation S230, that is, inputting the species feature tensor and the molecular feature tensor into the interaction layer to obtain a combined feature tensor of the species feature tensor and the molecular feature tensor, specifically includes operation S231 and operation S232.

[0105] In operation S231, the species feature tensor and the molecule feature tensor are dot-multiplied to obtain a mixed feature tensor.

[0106] In operation S232, the species feature tensor, the mixed feature tensor, and the molecular feature tensor are concatenated to obtain a combined feature tensor.

[0107] According to an embodiment of the present disclosure, before performing dot product on the species feature tensor and the molecule feature tensor, the method further includes converting the species feature tensor and the molecule feature tensor into two feature tensors with the same shape through a fully connected layer respectively.

[0108] Specifically, the molecular feature tensor A can be converted into a tensor A' of a specific length L through a fully connected layer, and the species feature tensor B can be converted into a tensor B' of a specific length L through a fully connected layer. A' and B' are dot-multiplied to obtain tensor C, and A', C and B' are connected in sequence to obtain the final combined feature tensor D. The combined feature tensor D is input into the prediction layer to obtain the acute toxicity prediction result.

[0109] According to the embodiments of the present disclosure, the shapes of the molecular feature tensor A and the species feature tensor B can be transformed to be consistent and then dot-multiplied and spliced, so that the molecular features and species information of the compound can be combined, and then subsequent feature extraction can be performed based on the combined features. Further, for ease of understanding, the above-mentioned transformation dot-multiplication and splicing operations are illustrated as follows:

[0110] The molecular feature tensor A is represented as ([[1., 1., 1.],

[0111] [0., 0., 0.]]).

[0112] The shape of the numerator feature tensor is (2,3).

[0113] First, the molecular feature tensor A is converted into a one-dimensional vector X1 with a length of 3, and the species information feature tensor B is converted into a one-dimensional vector Y1 with a length of 3.

[0114] At this time, the two feature tensors X1 and Y1 have the same first two dimensions, and can be dot-multiplied to obtain the tensor Z, where: the feature tensor X1 is represented as [1, 0, 1], the feature tensor Y1 is represented as [3, 4, 0], the feature tensor Z is represented as [3, 0, 0], and the concatenated feature tensor D is represented as [1, 0, 1, 3, 0, 0, 3, 4, 0].

[0115] In order to better illustrate the composition of toxicity data, taking the acute toxicity data of volatile organic compound isoprene to zebrafish as an example, the toxicity data composition is compound SMILES: C=CC(=C)C, zebrafish Cytb gene sequence: GCTTATCC (partial sequence), toxicity value: 12 mg / L, kingdom: Animalia, phylum: Chordata, class: Osteichthyes, order: Cypriniformes, family: Cyprinidae, genus: Danio, species: zebrafish.

[0116] For example, SMILES encoding is a unique string obtained by mapping compounds one by one, reflecting the structure of the compound. The common sequence obtained after matching and aligning the Cytb sequence is a string of length 1041 containing four codons ATCG. According to A-[1,0,0,0], T-[0,1,0,0], C-[0,0,1,0], G-[0,0,0,1], the one-hot encoding of the Cytb sequence (part) GCTTATCC is obtained, and the encoding is the following [[0,0,0,1],[0,0,1,0],[0,1,0,0],[0,1,0,0],[1,0,0,0],[0,1,0,0],[0,0,1,0],[0,0,1,0]] matrix. Gather all the biological evolutionary populations of the species involved and represent them using a string of length equal to their number. Assign 1 to the corresponding positions of the 7 evolutionary populations to which zebrafish belongs and 0 to the remaining positions.

[0117] According to an embodiment of the present disclosure, in operation S240, the prediction layer 330 may include a plurality of fully connected layers, each of which includes a discard layer to prevent overfitting of the initial prediction model.

[0118] According to an embodiment of the present disclosure, the combined feature tensor may be sequentially input into a plurality of fully connected layers to obtain initial acute toxicity prediction data.

[0119] According to an embodiment of the present disclosure, the discard layer is a dropout layer, and the coefficient may be P. During the training process, a part of neurons is randomly turned off and temporarily removed from the network with a certain probability, thereby preventing the model from being overly dependent on certain specific neurons or features, reducing the mutual dependence between neurons, and improving the robustness of the model.

[0120] According to an embodiment of the present disclosure, each node of the current layer in the fully connected layer is connected to all the nodes of the previous layer, and the number of nodes in each layer is fA, fB, 1 respectively. Except for the last layer, the output of each layer uses the ReLU activation function (Linear rectification function) to convert the linear features in the neural network into nonlinear features; wherein the last layer of the fully connected layer is the output layer, which outputs the initial acute toxicity data of the corresponding compound to the aquatic organisms of the species.

[0121] Figure 5 A flowchart of a method for establishing a multi-species aquatic organism acute toxicity prediction model according to another embodiment of the present disclosure is schematically shown.

[0122] like Figure 5 As shown, the method for establishing a multi-species aquatic organism acute toxicity prediction model includes operations S501~S507.

[0123] In operation S501, actual acute toxicity data of a plurality of compounds to a plurality of species of aquatic organisms, and species information of a plurality of compounds and a plurality of species of aquatic organisms are obtained.

[0124] In operation S502, a molecular graph of the compound is drawn according to SMILES. The specific drawing method is referred to the above embodiment and will not be described in detail.

[0125] In operation S503, the species information is encoded. The specific encoding method is referred to the above embodiment and will not be described in detail.

[0126] In operation S504 , the data set is divided into a training set, a validation set, and a test set.

[0127] In operation S505, the initial prediction model is trained and parameters are adjusted. Specifically, the network parameters of the initial prediction model are optimized using the data of the training set and the validation set.

[0128] According to an embodiment of the present disclosure, in operation S300 and operation S505, the network parameters of the initial prediction model may include model parameters and model hyperparameters. In order to determine the optimal prediction model, the actual acute toxicity data of multiple compounds on multiple species of aquatic organisms, the molecular graphs of multiple compounds, and the species information of multiple species of aquatic organisms are divided into a training set and a validation set. Operation S300 and operation S505 specifically include operation S310 to sub-step S340.

[0129] In operation 310, based on a set of preset values ​​of model hyperparameter combinations, the graph neural network model is iteratively trained using the reaction rate constant data and the reaction rate constant prediction data in the training set, and the model parameters are optimized to obtain a trained graph neural network model.

[0130] In operation 320 , the trained graph neural network model is used to predict the validation set to obtain statistical parameters.

[0131] In operation 330, based on the automatic parameter adjustment algorithm, the preset values ​​of the model hyperparameter combination are adjusted, and the iterative training and prediction operations are repeated until the optimal statistical parameters are determined, wherein a set of model hyperparameter combinations corresponding to the optimal statistical parameters are the optimal model hyperparameters.

[0132] In operation 340 , the trained graph neural network model obtained based on the optimal model hyperparameters is determined as a multi-species aquatic organism acute toxicity prediction model of the compound to multi-species aquatic organisms.

[0133] According to an embodiment of the present disclosure, the model hyperparameter combination is ,in, is the learning rate, which is used to control the progress of convergence to the local minimum; is the weight decay parameter, which is used to reduce the complexity of the model; P is the random inactivation ratio of neurons in the fully connected layer, which is used to prevent overfitting of the model; S is the batch processing data size; T is the iteration step size in the graph isomorphic network.

[0134] According to the embodiments of the present disclosure, in order to improve the model prediction effect, the automatic parameter adjustment algorithm is preferably a Bayesian optimization algorithm; the statistical parameters include at least one of the mean square error (MSE), the root mean square error (RMSE), the mean absolute error (MAE), and the coefficient of determination (R2). The model parameters corresponding to the lowest RMSE of the validation set result in the E generation (epaoch) iteration are selected and saved.

[0135] In operation S506, the initial prediction model is evaluated. Specifically, the initial prediction model is evaluated using the data of the test set.

[0136] According to an embodiment of the present disclosure, actual acute toxicity data of multiple compounds on multiple species of aquatic organisms, molecular graphs of the multiple compounds and species information of the multiple species of aquatic organisms are also divided into test sets. The establishment method of the present disclosure also includes: using the test set to evaluate the performance of the multi-species aquatic organism acute toxicity prediction model.

[0137] In operation S507, a trained multi-species aquatic organism acute toxicity prediction model is obtained.

[0138] Figure 6 The flowchart of the acute toxicity prediction method according to the embodiment of the present disclosure is schematically shown.

[0139] like Figure 6 As shown, the method includes operations S610 to S620.

[0140] In operation S610, a molecular graph of a compound to be predicted and species information of an organism to be predicted are obtained, wherein the species information includes a Cytb common subsequence and / or species evolution information of the organism to be predicted.

[0141] In operation S620, the molecular graph of the compound to be predicted and the species information of the organism to be predicted are input into a multi-species aquatic organism acute toxicity prediction model to obtain acute toxicity prediction data of the compound to be predicted on the organism to be predicted; wherein, the multi-species aquatic organism acute toxicity prediction model is obtained by establishing the method described above.

[0142] The acute toxicity prediction method provided in the embodiments of the present disclosure is suitable for predicting the acute toxicity of large-scale compounds and multi-species aquatic organisms. The prediction method is simple and efficient, improves the prediction performance, expands the prediction range, and has broad application prospects in the fields of chemical environmental health risk assessment, environmental assessment, etc.

[0143] The acute toxicity prediction method provided by the embodiments of the present disclosure can more reliably predict the acute toxicity of the compound to be predicted to various aquatic organisms.

[0144] Figure 7 The block diagram of the device for establishing the multi-species aquatic organism acute toxicity prediction model according to the embodiment of the present disclosure is schematically shown.

[0145] like Figure 7 As shown, the device 700 for establishing a multi-species aquatic organism acute toxicity prediction model includes a first acquisition module 710 , a first obtaining module 720 , and an optimization module 730 .

[0146] The first acquisition module 710 is used to obtain actual acute toxicity data of multiple compounds on multiple species of aquatic organisms, molecular graphs of multiple compounds and species information of multiple species of aquatic organisms, wherein the species information includes Cytb common subsequences and / or species evolution information of multiple species of aquatic organisms.

[0147] The first acquisition module 720 is used to input the molecular graphs of multiple compounds and the species information of multiple species of aquatic organisms into the initial prediction model to obtain initial acute toxicity prediction data of the multiple compounds to the multiple species of aquatic organisms.

[0148] The optimization module 730 is used to optimize the network parameters of the initial prediction model according to the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

[0149] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging circuits, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.

[0150] For example, any of the first acquisition module 710, the first acquisition module 720, and the optimization module 730 can be combined in one module / unit / sub-unit for implementation, or any of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the first acquisition module 710, the first acquisition module 720, and the optimization module 730 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the first acquisition module 710, the first obtaining module 720, and the optimization module 730 may be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function may be executed.

[0151] Figure 8 The block diagram of the device for establishing the multi-species aquatic organism acute toxicity prediction model according to the embodiment of the present disclosure is schematically shown.

[0152] like Figure 8 As shown, the device 700 for establishing a multi-species aquatic organism acute toxicity prediction model includes a second acquisition module 810 and a second obtaining module 820 .

[0153] The second acquisition module 810 is used to acquire the molecular graph of the compound to be predicted and the species information of the organism to be predicted, wherein the species information includes the Cytb common subsequence and / or species evolution information of the organism to be predicted.

[0154] The second acquisition module 820 is used to input the molecular graph of the compound to be predicted and the species information of the organism to be predicted into the multi-species aquatic organism acute toxicity prediction model to obtain the acute toxicity prediction data of the compound to be predicted to the organism to be predicted; wherein, the multi-species aquatic organism acute toxicity prediction model is obtained by a method establishment method.

[0155] It should be noted that the data processing system part in the embodiments of the present disclosure corresponds to the data processing method part in the embodiments of the present disclosure. The description of the data processing system part specifically refers to the data processing method part, which will not be repeated here.

[0156] Fig. 9 A block diagram of an electronic device suitable for implementing the method described above according to an embodiment of the present disclosure is schematically shown. Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0157] like Fig. 9 As shown, the electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0158] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0159] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the input / output (I / O) interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0160] According to an embodiment of the present disclosure, the method flow according to an embodiment of the present disclosure can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed. According to an embodiment of the present disclosure, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0161] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0162] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0163] For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 902 and / or the RAM 903 described above and / or one or more memories other than the ROM 902 and the RAM 903 .

[0164] The embodiments of the present disclosure also include a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiments of the present disclosure. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the method for establishing a multi-species aquatic organism acute toxicity prediction model or the toxicity prediction method provided by the embodiments of the present disclosure.

[0165] When the computer program is executed by the processor 901, the above functions defined in the system / device of the embodiment of the present disclosure are executed. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0166] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0167] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0168] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0169] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for establishing a multi-species aquatic organism acute toxicity prediction model, comprising: Acquiring actual acute toxicity data of multiple compounds on multiple species of aquatic organisms, molecular graphs of the multiple compounds, and species information of the multiple species of aquatic organisms, wherein the species information includes Cytb common subsequences and / or species evolution information of the multiple species of aquatic organisms; Inputting the molecular graphs of the plurality of compounds and the species information of the plurality of aquatic organisms into an initial prediction model to obtain initial acute toxicity prediction data of the plurality of compounds to the plurality of aquatic organisms; The network parameters of the initial prediction model are optimized according to the initial acute toxicity prediction data and the actual acute toxicity data to obtain a trained multi-species aquatic organism acute toxicity prediction model.

2. The method according to claim 1, wherein: The initial prediction model includes a feature extraction network, an interaction layer and a prediction layer connected in sequence, wherein the feature extraction network includes a species feature extraction network and a molecular feature extraction network; The step of inputting the molecular graphs of the plurality of compounds and the species information of the plurality of aquatic organisms into the initial prediction model to obtain the initial acute toxicity prediction data of the plurality of compounds to the plurality of aquatic organisms comprises: Inputting the species information of the multi-species aquatic organisms into the species feature extraction network to extract the species feature tensor of the species information of the multi-species aquatic organisms; Inputting the molecular graphs of the plurality of compounds into the molecular feature extraction network to extract molecular feature tensors of the molecular graphs of the plurality of compounds; Inputting the species feature tensor and the molecule feature tensor into the interaction layer to obtain a combined feature tensor of the species feature tensor and the molecule feature tensor; The combined feature tensor is input into the prediction layer to obtain the initial acute toxicity prediction data.

3. The method according to claim 2, wherein: The step of inputting the species feature tensor and the molecule feature tensor into the interaction layer to obtain a combined feature tensor of the species feature tensor and the molecule feature tensor comprises: Performing a dot product of the species feature tensor and the molecule feature tensor to obtain a mixed feature tensor; The species feature tensor, the mixed feature tensor and the molecular feature tensor are concatenated to obtain the combined feature tensor.

4. The method according to claim 2, wherein: The species information includes the Cytb common subsequences and species evolution information of the multiple species of aquatic organisms; The species feature tensor for extracting species information of the multi-species aquatic organisms includes: Performing feature extraction on the Cytb common subsequence to obtain a Cytb feature tensor; Extracting features of the species evolution information to obtain a species evolution tensor; The Cytb feature tensor and the species evolution tensor are concatenated to obtain the species feature tensor.

5. The method according to claim 4, wherein: The extracting features of the Cytb common subsequence to obtain a Cytb feature tensor comprises: One-hot encoding the Cytb common subsequences of the multi-species aquatic organisms according to the base order to obtain a Cytb feature tensor; A convolution kernel is used to perform feature extraction on the Cytb feature tensor to obtain the Cytb feature tensor.

6. The method according to claim 4, wherein: The species evolution information includes species classification information of the multi-species aquatic organisms; The feature extraction of the species evolution information to obtain the species evolution tensor includes: One-hot encoding is performed on the species classification information to obtain a species evolution tensor; A fully connected layer is used to extract features of the species evolution tensor to obtain the species evolution tensor.

7. The method according to claim 2, wherein: The molecular feature extraction network is a graph isomorphic network.

8. The method of establishing according to claim 2, wherein: The prediction layer includes a plurality of fully connected layers, each of which includes a discard layer for preventing the initial prediction model from overfitting.

9. The method according to claim 1, wherein: The molecular graphs of the plurality of compounds include node features for representing atomic features and edge features for representing bond features; The molecular graphs of the various compounds are constructed by the following method: According to the SMILES codes of the plurality of compounds, adding the node features for characterizing the atomic feature information and the edge features for characterizing the bond feature information to an undirected graph to obtain a molecular graph of the plurality of compounds; The atomic characteristic information includes one or more of atomic symbol type, atomic implicit valence, atomic free electron number, atomic connection number, atomic formal charge, atomic hybridization mode, atomic connection hydrogen atom number, atomic chirality, and FCFP fingerprint characteristics; The bond characteristic information includes one or more of the bond type, whether the bond is conjugated, whether the bond is on a ring, and the bond stereo configuration information.

10. A method for predicting acute toxicity, comprising: Acquire a molecular graph of the compound to be predicted and species information of the organism to be predicted, wherein the species information includes a Cytb common subsequence and / or species evolution information of the organism to be predicted; Inputting the molecular graph of the compound to be predicted and the species information of the organism to be predicted into a multi-species aquatic organism acute toxicity prediction model to obtain acute toxicity prediction data of the compound to be predicted to the organism to be predicted; Wherein, the multi-species aquatic organism acute toxicity prediction model is obtained by the establishment method described in any one of claims 1 to 9.

Images